VBVR-Pro Dataset
300 Tasks
A large-scale, multi-modal synthetic dataset comprising 300 tasks across diverse visual environments and reasoning capabilities. The dataset supports multiple aligned visual modalities, including standalone images, interleaved text–image inputs, and videos, enabling systematic evaluation across different forms of visual information. All tasks are organized into five categories: perception, spatiality, transformation, abstraction, and knowledge. A single task may be associated with multiple taxonomy tags, represented by multiple taxonomy-specific colors in its task icon.
Download
Knowledge
87 tasks
Abstraction
85 tasks
Spatiality
47 tasks
Transformation
40 tasks
Perception
70 tasks
Communicating Vessel Level Equalization
Image sequence
SFT-Image source
Image prompt
Two connected vertical tubes contain a liquid. The two liquid levels change until both tubes reach the same equilibrium level. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the liquid levels changing from the initial state to the shared equilibrium level at uniform stages.
Video
Data video source
Video prompt
Two connected vertical tubes contain a liquid. The two liquid levels slowly and smoothly change until both tubes reach the same equilibrium level, then the scene remains still.
Add Matchstick Make Three Squares
Image sequence
SFT-Image source
Image prompt
Add exactly 1 matchstick so the layout gains the maximum possible completed 1x1 squares on this grid. Final layout target: 8 completed unit squares. Output 1 image showing the final completed matchstick layout.
Video
Data video source
Video prompt
Add exactly 1 matchstick so the layout gains the maximum possible completed 1×1 squares on this grid.
Best Time To Buy And Sell Stock
Image sequence
SFT-Image source
Image prompt
The image shows a stock price series over consecutive days (each point's value is labeled). You may make a single transaction: buy on one day and sell on one later day. Compute the maximum profit achievable (the sell-day price minus the buy-day price). If no profitable trade exists (prices never rise after a low), the answer is 0. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows a stock price series over consecutive days (each point's value is labeled). You may make a single transaction: buy on one day and sell on one later day. Compute the maximum profit achievable (the sell-day price minus the buy-day price). If no profitable trade exists (prices never rise after a low), the answer is 0. Generate a video that marks the best buy and sell days on the chart and reveals the maximum profit. Only modify what the task requires; keep the background and any other element unchanged.
Domino Chain Gap Analysis
Image sequence
SFT-Image source
Image prompt
Analyze the domino chain to find which domino is the last to fall. Push the first domino so that each domino falls in sequence. The chain stops at the gap where the spacing between two upright dominos is visibly wider than the others. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images at uniform time intervals during the collapse; the last image shows the final settled state.
Video
Data video source
Video prompt
Analyze the domino chain to find which domino is the last to fall. Push the first domino and watch as each domino falls in sequence. The chain stops at the gap where the spacing between two upright dominos is visibly wider than the others. Only modify what the task requires; keep the background and any other element unchanged.
Wood Slide
Image sequence
SFT-Image source
Image prompt
Shown is a wooden sliding-block puzzle. Each piece can be pushed up, down, left, or right into adjacent empty squares, but never lifted out, turned, or overlapped with another piece. Maneuver the marked target block to the dashed goal cell. Output 1 image showing the board after each move, in order. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Shown is a wooden sliding-block puzzle. Each piece can be pushed up, down, left, or right into adjacent empty squares, but never lifted out, turned, or overlapped with another piece. Maneuver the marked target block to the dashed goal cell. Generate a video that animates a sequence of slides, moving one block at a time, and ends with the target seated on the goal. Only modify what the task requires; keep the background and any other element unchanged.
Isosceles Trapezoid Fourth Vertex
Image sequence
SFT-Image source
Image prompt
Two thick black segments meet at one vertex and are two consecutive sides of a quadrilateral. Exactly one pair of opposite sides must be parallel and the two non-parallel sides must be equal in length (an isosceles trapezoid). Five lettered circles A-E mark candidate fourth vertices to the left. Find the fourth vertex that completes the isosceles trapezoid, with that vertex marked red. Output 1 image showing the final figure with the correct fourth vertex marked red.
Video
Data video source
Video prompt
You see two joined sides of a quadrilateral and five choices A through E. Complete the shape as an isosceles trapezoid: one pair of opposite sides parallel, legs equal. Pick the circle that is the true fourth vertex. Animate so only the correct choice turns red at the end.
Parallelogram Fourth Vertex
Image sequence
SFT-Image source
Image prompt
Complete the parallelogram from the two given adjacent sides by drawing the two missing sides and keeping them visible; then highlight in red the correct fourth-vertex option. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final figure with the parallelogram completed and the correct fourth vertex marked red.
Video
Data video source
Video prompt
Complete the parallelogram from the two given adjacent sides by drawing the two missing sides and keeping them visible; then highlight in red the correct fourth-vertex option. Only modify what the task requires; keep the background and any other element unchanged.
Circle Midpoint Identification
Image sequence
SFT-Image source
Image prompt
Two large empty circles and five smaller labeled circles (A-E) appear on a white canvas. A straight segment connects the centers of the two large circles. Exactly one small circle lies at the true geometric midpoint of that segment; it should be filled red. Keep all circles separated with no overlapping disks. Output 1 image showing the final canvas with the centerline drawn and the midpoint circle red.
Video
Data video source
Video prompt
Starting from the given arrangement, animate a full solution: add the centerline between the two big circles, then turn the midpoint option red. Maintain non-overlapping geometry and a square 1024×1024 composition throughout.
Understand Scene Structure
Image sequence
SFT-Image source
Image prompt
The scene shows a floorplan with multiple rooms. Identify the bathroom and draw a green rectangular box around it to highlight it. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
The scene shows a floorplan with multiple rooms. Identify the bathroom and draw a green rectangular box around it to highlight it. Only modify what the task requires; keep the background and any other element unchanged.
Triangle Circumcenter Identification
Image sequence
SFT-Image source
Image prompt
The image shows a triangle and five small hollow circular markers placed nearby. Each marker stands for one candidate location. Exactly one marker is the circumcenter of the triangle (the common intersection point of the perpendicular bisectors of the three sides). Determine the single true circumcenter marker, with it singled out on screen (for example with a highlight or fill), without labeling options by number or word. Output 1 image showing the final figure with the circumcenter marker singled out.
Video
Data video source
Video prompt
The image shows a triangle and five small hollow circular markers placed nearby. Each marker stands for one candidate location. Exactly one marker is the circumcenter of the triangle (the common intersection point of the perpendicular bisectors of the three sides). Which of the five markers lies at the circumcenter of the triangle? Show it in motion. Generate a single continuous video. Continue from this frame by visually singling out the true circumcenter only (for example with a brief highlight, fill, pulse, or gentle zoom), without labeling options by number or word.
Perpendicular Bisector Option Identification
Image sequence
SFT-Image source
Image prompt
Using the existing segment between the two small circles, draw its perpendicular bisector, then highlight in red the option lying on that perpendicular bisector. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result with the perpendicular bisector drawn and the correct option highlighted in red.
Video
Data video source
Video prompt
From the existing segment between the two small circles, draw the perpendicular bisector, retain the construction, and highlight in red the matching option.
Triangle Incenter Identification
Image sequence
SFT-Image source
Image prompt
Identify which labeled point is the triangle's incenter and color that marker red. The incenter is the point where the three internal angle bisectors meet; it is also the center of the inscribed circle. Among the five lettered points inside the triangle, pick exactly one label and leave all other markers unchanged. Output 1 image showing the final figure with the true incenter marked red.
Video
Data video source
Video prompt
Highlight the true incenter among options A–E by filling its circle red. Use angle-bisector geometry: the incenter is inside the triangle where bisectors concur.
Square Fifth Point Identification
Image sequence
SFT-Image source
Image prompt
Among the five markers, four form a square and one does not; highlight in red only the marker that is not a vertex of that square. Output 1 image showing the final result with the outlier marker highlighted in red.
Video
Data video source
Video prompt
Among the five markers, four form a square and one does not; highlight in red only the marker that is not a vertex of that square.
Chart Extreme With Data
Image sequence
SFT-Image source
Image prompt
The scene shows a area chart with numeric data labels. Find the maximum value point and draw a red rectangular border around the corresponding point to highlight it. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
The scene shows a area chart with numeric data labels. Find the maximum value point and draw a red rectangular border around the corresponding point to highlight it. Only modify what the task requires; keep the background and any other element unchanged.
Mario Hit Question Block
Image sequence
SFT-Image source
Image prompt
An NES-style Mario scene where Mario bumps the question block from below, turning it into a used block and releasing a coin that pops out and disappears. Only modify what the task requires; keep the background and any other element unchanged. Output 3 images of the key moments: Mario approaching the block, reaching it from below, and bumping it so the coin pops out as he lands.
Video
Data video source
Video prompt
Mario approaches, jumps, and bumps the question block, spawning a coin that disappears as the block becomes used — generate this pixel-art interaction. Only modify what the task requires; keep the background and any other element unchanged.
Angle Bisector Ball Identification
Image sequence
SFT-Image source
Image prompt
Draw the angle bisector starting from the vertex so it splits the angle into two equal parts. Then indicate which labeled ball (A–E) lies exactly on the bisector by marking that ball solid red; keep all other balls white with black outlines. Use black lines on white, and ensure no circles or labels overlap. Output 1 image showing the final diagram with the bisector drawn and the correct ball marked red.
Video
Data video source
Video prompt
Draw the angle bisector starting from the vertex so it splits the angle into two equal parts. Animate the bisector appearing from the vertex outward. Then indicate which labeled ball (A–E) lies exactly on the bisector by marking that ball solid red; keep all other balls white with black outlines. Use black lines on white, and ensure no circles or labels overlap.
Construction Stack
Image sequence
Data image source
Image prompt
Match the left CURRENT stacks to the TARGET on the right by moving blocks (one top block per move). Keep the TARGET stacks on the right unchanged. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each block move, showing the updated stack state.
Video
Data video source
Video prompt
Match the left CURRENT stacks to the TARGET on the right by moving blocks (one top block per move). Keep the TARGET stacks on the right unchanged. Only modify what the task requires; keep the background and any other element unchanged.
Circle Center Identification
Image sequence
SFT-Image source
Image prompt
A large circle is drawn on white. Five small lettered circles (A-E) lie inside it; exactly one marks the geometric center of the large circle. Determine that center marker, with it filled solid red while the other markers remain unchanged. Output 1 image showing the final diagram with the center marker filled red.
Video
Data video source
Video prompt
A large circle holds five small lettered markers A–E; one of them coincides with the center of the large circle. Show the full progression from the start frame to the state where only the true center marker turns red, then hold it.
Tangent Point Identification
Image sequence
SFT-Image source
Image prompt
A large circle and an external point are shown. Five lettered options (A through E) lie on the circle's circumference. Determine which labeled point on the circle is the tangency point for a line drawn from the external point so that it just touches the circle, with only that option's marker filled red. Output 1 image showing the final diagram with the correct tangency marker red.
Video
Data video source
Video prompt
A large circle and an external point are shown. Five lettered options (A through E) lie on the circle’s circumference. Animate the tangent segment from the external point out to the true tangency on the rim, then mark that option red.
Sudoku Missing Number Completion
Image sequence
SFT-Image source
Image prompt
Complete this 9x9 Sudoku by filling every blank cell with the digit uniquely determined by Sudoku constraints on its row, column, and 3x3 box, until the grid is fully solved. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the fully solved grid.
Video
Data video source
Video prompt
Solve the Sudoku step by step using this exact rule: at each step, fill the topmost-leftmost empty cell whose value is uniquely determined by Sudoku constraints. Repeat until no blanks remain. Only modify what the task requires; keep the background and any other element unchanged.
Gomoku Five In A Row Completion
Image sequence
SFT-Image source
Image prompt
Gomoku rules: players alternate placing stones on empty intersections of the board, and the first to make five consecutive stones in a straight line (horizontal, vertical, or diagonal) wins. It is the current side's turn. Place one stone to complete an immediate five-in-a-row win. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board after placing the winning stone.
Video
Data video source
Video prompt
Gomoku rules: players alternate placing stones on empty intersections of the board, and the first to make five consecutive stones in a straight line (horizontal, vertical, or diagonal) wins. It is the current side's turn. Place one stone to complete an immediate five-in-a-row win. Only modify what the task requires; keep the background and any other element unchanged.
Circle Largest Numerical Value
Image sequence
Data image source
Image prompt
The scene shows 6 numbers on a white canvas. First compare the numerical values of all numbers, then draw one red circle around the single largest number. Do not circle any other numbers. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
The scene shows 6 numbers on a white canvas. First compare the numerical values of all numbers, then draw one red circle around the single largest number. Do not circle any other numbers. Show the complete circling process step by step. Only modify what the task requires; keep the background and any other element unchanged.
Tic-Tac-Toe Winning Move Prediction
Image sequence
SFT-Image source
Image prompt
Tic-Tac-Toe position with exactly one immediate winning move for the side to play. Place that mark in the winning empty cell. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board after placing the winning mark.
Video
Data video source
Video prompt
Tic-Tac-Toe position with exactly one immediate winning move for the side to play. Place that mark in the winning empty cell. Only modify what the task requires; keep the background and any other element unchanged.
Spider Solitaire Stack Merge
Image sequence
SFT-Image source
Image prompt
A one-suit Spider Solitaire tableau on green felt: several columns of face-down backs and face-up spades. Make the single needed stack move: move a legal descending run of spades onto another column so the bottom card of the run sits on the face-up card that is exactly one rank higher, extending the sequence. Show the resulting tableau after the merge. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the tableau after the stack merge.
Video
Data video source
Video prompt
Spider Solitaire (1 suit): make the valid move that moves a tail of face-up cards onto another pile whose exposed bottom card is exactly one rank above the bottom card of the moved segment. Animate the tail sliding onto its destination pile. Only modify what the task requires; keep the background and any other element unchanged.
Ultra Tic-Tac-Toe Best Next Move
Image sequence
SFT-Image source
Image prompt
An Ultimate Tic-Tac-Toe board is shown with the current player marked, and the single best legal move played for that player: the forced local board is outlined in orange and the chosen cell is marked with a green box. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board after the best move is played.
Video
Data video source
Video prompt
An Ultimate Tic-Tac-Toe board is shown with the current player marked at the top. Generate a video that plays the single best legal move for that player, respecting the forced-board rule (the local board you must play in is outlined in orange). Starting from this board, it shows the legal candidate cells as small green dots, may box reviewed candidates in orange, then marks the final chosen cell with a green box and places the move there. The video ends right after that move is placed; the camera is fixed and there is no later destination-board or next-turn animation. Only modify what the task requires; keep the background and any other element unchanged.
Freecell Foundation Progress Move
Image sequence
SFT-Image source
Image prompt
On this FreeCell tableau (four free cells and four foundation piles on top, eight columns below), make the single move that most directly advances a card onto a foundation pile while keeping the position legal and sensible: prefer moving the lowest eligible rank to foundations when several are possible, and prefer clearing a tableau column over moving the same card from a free cell. Show the resulting board after the move. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the board after the foundation move.
Video
Data video source
Video prompt
From the current FreeCell layout, make the one foundation move that best increases foundation progress without breaking rules. Among legal moves, favor the smallest rank that may go up (aces first), then favor taking it from the bottom of a column rather than from a free cell if both exist. Animate that card sliding onto its foundation. Only modify what the task requires; keep the background and any other element unchanged.
Gomoku Best Next Move
Image sequence
SFT-Image source
Image prompt
Gomoku rules: players alternate placing stones on empty cells of the board, and the first to make five consecutive stones in a straight line (horizontal, vertical, or diagonal) wins. It is white's turn. Place one white stone at the best next cell. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board after placing the white stone.
Video
Data video source
Video prompt
Gomoku rules: players alternate placing stones on empty cells of the board, and the first to make five consecutive stones in a straight line (horizontal, vertical, or diagonal) wins. It is white's turn. Place one white stone at the best next cell. Only modify what the task requires; keep the background and any other element unchanged.
Word Search Start End Coordinate
Image sequence
SFT-Image source
Image prompt
A word-search panel shows a letter grid with numbered row and column labels and a target word in a badge, with the hidden word marked along its single straight line: the start cell outlined green, a yellow line through the word's letters, and the end cell outlined red, establishing the word's start cell, end cell, and direction. Output 1 image showing the grid with the found word marked.
Video
Data video source
Video prompt
A word-search panel shows a letter grid with numbered row and column labels and a target word in a badge. The target word is hidden along one straight line (horizontal, vertical, or diagonal). Generate a video that finds and traces it: outline the word's start cell in green, sweep a yellow line along the straight path through its letters, and outline the end cell in red. The camera is fixed; only the tracing highlight animates, establishing the word's start cell, end cell, and direction.
Tangent Line Option Identification
Image sequence
SFT-Image source
Image prompt
A large circle has a marked point on its boundary. Five labeled options A-E appear nearby. Consider the tangent line to the circle at that point, extending outward. Determine the single option whose center lies on that tangent line, with it filled or outlined in red. Output 1 image showing the final diagram with the correct option marked red.
Video
Data video source
Video prompt
A large circle has a marked point on its boundary. Five labeled options A–E appear nearby. Imagine the tangent line to the circle at that point, extending outward. Animate so only the single option whose center lies on that tangent line is filled or outlined in red at the end.
High Density Liquid
Image sequence
Data image source
Image prompt
There are 3 cups of liquid; above each cup is one identical object (same mass and density), which will fall into the corresponding cup. One cup has a different liquid color from the others, and that cup's liquid color is darker; that cup's liquid density is higher than the object, and the others are lower. By buoyancy: when the liquid density is higher than the object, the object floats; when lower, it sinks. Show where each object should be in each cup and the final floating or sunk state. Only modify what the task requires; keep the background and any other element unchanged. Output 3 images: objects mid-fall between start and water surface, objects at the water center, and the final floating or sunk state.
Video
Data video source
Video prompt
There are 3 cups of liquid; above each cup is one identical object (same mass and density), which will fall into the corresponding cup. One cup has a different liquid color from the others, and that cup's liquid color is darker; that cup's liquid density is higher than the object, and the others are lower. By buoyancy: when the liquid density is higher than the object, the object floats; when lower, it sinks. Show where each object should be in each cup and the final floating or sunk state. Only modify what the task requires; keep the background and any other element unchanged.
Multiple Occlusions Horizontal
Image sequence
SFT-Image source
Image prompt
The scene shows 4 objects arranged horizontally on the right side of the frame, with a dark rectangular mask initially positioned on the left side. Move the mask horizontally to the right in a continuous motion until it leaves the frame. As it moves, the mask passes in front of the objects, temporarily blocking them from view. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the mask at uniform horizontal progress.
Video
Data video source
Video prompt
The scene shows 4 objects arranged horizontally on the right side of the frame, with a dark rectangular mask initially positioned on the left side. Move the mask horizontally to the right in a continuous motion until it leaves the frame. As it moves, the mask passes in front of the objects, temporarily blocking them from view. Only modify what the task requires; keep the background and any other element unchanged.
Minesweeper Safe Cell Inference
Image sequence
SFT-Image source
Image prompt
In Minesweeper, every revealed number tells how many mines lie among that cell's 8 surrounding neighbours. A flag marks a cell already known to hold a mine, and grey cells are unrevealed. Reasoning from the numbers and flags alone: a hidden cell is GUARANTEED SAFE if it cannot hold a mine (e.g. once a number's mines are all accounted for by adjacent flags, its other hidden neighbours must be safe), and GUARANTEED A MINE if it must hold one (e.g. when a number's remaining mine count equals how many hidden neighbours it has left, those neighbours are all mines). Find every unrevealed cell that is guaranteed safe and outline it in green. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image of the board with every guaranteed-safe cell outlined green.
Video
Data video source
Video prompt
In Minesweeper, every revealed number tells how many mines lie among that cell's 8 surrounding neighbours. A flag marks a cell already known to hold a mine, and grey cells are unrevealed. Reasoning from the numbers and flags alone: a hidden cell is GUARANTEED SAFE if it cannot hold a mine (e.g. once a number's mines are all accounted for by adjacent flags, its other hidden neighbours must be safe), and GUARANTEED A MINE if it must hold one (e.g. when a number's remaining mine count equals how many hidden neighbours it has left, those neighbours are all mines). Deduce all the cells that are certainly safe to click next and outline them in green. Only modify what the task requires; keep the background and any other element unchanged.
Zuma Color Match Elimination
Image sequence
SFT-Image source
Image prompt
On the Archimedean spiral track, colored marbles advance toward the central hole. The triangle at the center holds your current shot color. Insert this ball into the gap along the chain that creates the longest same-color run and clears the largest contiguous group (three or more). Output one image showing the ball merged into the chosen gap with the matching run highlighted, then one image showing the chain after that run is eliminated. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Fire the ball into the insertion point on the moving chain that completes the strongest same-color match. Only one gap maximizes the eliminated run; animate the cannon rotating toward that gap, the projectile path, and the cleared marbles. Only modify what the task requires; keep the background and any other element unchanged.
Snake Future Body Position Prediction
Image sequence
SFT-Image source
Image prompt
Given the initial Snake board and move sequence, simulate the snake exactly and report the final head, final ordered body, collision status, and final board state. Only modify what the task requires; keep the background and any other element unchanged. Output 7 images showing the snake moving step by step at uniform stages.
Video
Data video source
Video prompt
Use the displayed Snake board as the starting state, apply every move in order, and answer with the complete final state including any collision details. Only modify what the task requires; keep the background and any other element unchanged.
Color Subtraction
Image sequence
SFT-Image source
Image prompt
Two colored balls move toward each other at identical speeds. Use subtractive color mixing in overlapping regions. In the final state both balls are fully merged at the midpoint. Use the standard subtractive multiply blend: mixed = round(c1 * c2 / 255) per channel (e.g. cyan * yellow = green). Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final merged state.
Video
Data video source
Video prompt
Animate two colored balls moving toward each other at identical speeds. Use subtractive color mixing in overlapping regions. Stop the animation when both balls fully merge at the midpoint. Use the standard subtractive multiply blend: mixed = round(c1 * c2 / 255) per channel (e.g. cyan * yellow = green). Only modify what the task requires; keep the background and any other element unchanged.
Pigment Color Mixing Subtractive
Image sequence
Data image source
Image prompt
The scene shows two solid-colored circular pigment blobs on the left and right of a white canvas, and a black-bordered square mixing zone in the centre. In subtractive (multiply) pigment mixing, the mixed color is the per-channel product of the two pigments: mixed = round(c1 * c2 / 255) (white keeps a color unchanged, black absorbs everything; e.g. cyan * yellow = green). The left pigment is an RGB(98, 232, 7) colored pigment and the right pigment is an RGB(105, 92, 90) colored pigment. Compute the subtractive mix and fill the black-bordered mixing zone with the resulting color, keeping everything else unchanged. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the mixing zone filled with the mixed color.
Video
Data video source
Video prompt
The scene shows two solid-colored circular pigment blobs on the left and right of a white canvas, and a black-bordered square mixing zone in the centre. In subtractive (multiply) pigment mixing, the mixed color is the per-channel product of the two pigments: mixed = round(c1 * c2 / 255) (white keeps a color unchanged, black absorbs everything; e.g. cyan * yellow = green). The left pigment is an RGB(98, 232, 7) colored pigment and the right pigment is an RGB(105, 92, 90) colored pigment. Compute the subtractive mix and fill the black-bordered mixing zone with the resulting color, keeping everything else unchanged. Only modify what the task requires; keep the background and any other element unchanged.
Code Output Completion
Image sequence
SFT-Image source
Image prompt
Read the code and the given input, compute the result, and replace the '?' on the '# Output: ?' line with that exact result, keeping the '# Output:' label. Do not change any other source content. Output 1 image showing the final code with the placeholder replaced by the result.
Video
Data video source
Video prompt
Read the code and the given input, compute the result, and replace the '?' on the '# Output: ?' line with that exact result, keeping the '# Output:' label. Do not change any other source content.
Gold Miner Hook Grab
Image sequence
SFT-Image source
Image prompt
The hook starts from the shown hanging position, then turns toward the object with the highest value-to-total-time ratio and grabs it. Values are fixed by type: small gold = 50, large gold = 180, rock = 20, bag = 90. Total time is the sum of turning time, straight extension time until first contact, grab time, and type-dependent retraction time. Turning time is max(8, round(angle_change_from_straight_down * 0.18)). Extension time is max(1, round((hit_distance - 92) / 14)). Grab time is always 8 frames. Retraction times are fixed: small gold = 26 frames, large gold = 62, rock = 70, bag = 40. Ties are broken by smaller total time, then smaller angle change, then left-to-right position. Determine which object will be grabbed. Only modify what the task requires; keep the background and any other element unchanged. Output 2 images: the hook reaching and grabbing the chosen object, then that object retracted back up to the winch.
Video
Data video source
Video prompt
The hook starts from the shown hanging position, then turns toward the object with the highest value-to-total-time ratio and grabs it. Values are fixed by type: small gold = 50, large gold = 180, rock = 20, bag = 90. Total time is the sum of turning time, straight extension time until first contact, grab time, and type-dependent retraction time. Turning time is max(8, round(angle_change_from_straight_down * 0.18)). Extension time is max(1, round((hit_distance - 92) / 14)). Grab time is always 8 frames. Retraction times are fixed: small gold = 26 frames, large gold = 62, rock = 70, bag = 40. Ties are broken by smaller total time, then smaller angle change, then left-to-right position. Determine which object will be grabbed. Only modify what the task requires; keep the background and any other element unchanged.
Chart Extreme Without Data
Image sequence
SFT-Image source
Image prompt
The scene shows a pie chart. Find the maximum value segment and draw a red outline around the corresponding segment to highlight it. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The scene shows a pie chart. Find the maximum value segment and draw a red outline around the corresponding segment to highlight it. Only modify what the task requires; keep the background and any other element unchanged.
Traffic Light
Image sequence
SFT-Image source
Image prompt
This scene shows a crossroad with two traffic lights (East, West). Each light independently follows a 3-color cycle: Red (4s) -> Yellow (4s) -> Green (4s) -> Yellow (4s) -> Red. Currently: East light is green with 4s countdown, West light is green with 1s countdown. Simulate 5 seconds and show the final state of both traffic lights. Only modify what the task requires; keep the background and any other element unchanged. Output one image per second of the countdown (every time a number changes or a light transitions, write one image). The last image shows the final state.
Video
Data video source
Video prompt
This scene shows a crossroad with two traffic lights (East, West). Each light independently follows a 3-color cycle: Red (4s) -> Yellow (4s) -> Green (4s) -> Yellow (4s) -> Red. Currently: East light is green with 4s countdown, West light is green with 1s countdown. Simulate 5 seconds and show the final state of both traffic lights. Only modify what the task requires; keep the background and any other element unchanged.
Color Addition
Image sequence
SFT-Image source
Image prompt
Two circular balls with different colors are positioned at different locations. They are shown after moving toward each other until they have completely merged at the midpoint. During overlap and in the final merged state, combine their colors using additive light mixing (as if two beams of light were overlapping). Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final merged state.
Video
Data video source
Video prompt
Two circular balls with different colors are positioned at different locations. They move toward each other at the same speed until they completely merge at the midpoint. During overlap and in the final merged state, combine their colors using additive light mixing (as if two beams of light were overlapping). Only modify what the task requires; keep the background and any other element unchanged.
Triangle Fermat Point Identification
Image sequence
SFT-Image source
Image prompt
Apply the standard Fermat-point construction to the given triangle: outer equilateral triangles on each edge, then three lines from the new apices to the opposite corners, meeting at the Fermat point. Mark the matching label (A-E) red. Output 1 image showing the final figure with the correct option marked red.
Video
Data video source
Video prompt
Animate the construction of the triangle's Fermat point (the point minimizing the total distance to the three vertices). Draw an equilateral triangle outward on each side, then a segment from each new outer apex to the opposite original vertex; where the three segments meet is the Fermat point. End by filling the matching option among the five labeled circles A–E solid red, and hold the final frame.
Go Capture Move
Image sequence
SFT-Image source
Image prompt
Go rules: stones are placed on intersections; connected stones of one color form a group; a group is captured when all its liberties are removed. Black to move. Place one black stone that immediately captures the target white group. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board after the capturing move, with the captured white group removed.
Video
Data video source
Video prompt
Go rules: stones are placed on intersections; connected stones of one color form a group; a group is captured when all its liberties are removed. Black to move. Place one black stone that immediately captures the target white group. Only modify what the task requires; keep the background and any other element unchanged.
Mirror Reflection
Image sequence
SFT-Image source
Image prompt
Given the mirror reflectivity = 0.82, predict how light reflects when it encounters the mirror. Extend the reflected ray to the image boundary. The angle annotation should be removed. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the incident ray and the predicted red reflected ray.
Video
Data video source
Video prompt
Given the mirror reflectivity = 0.82, predict how light reflects when it encounters the mirror. Extend the reflected ray to the image boundary. The angle annotation should be removed. Only modify what the task requires; keep the background and any other element unchanged.
Math Answer Completion
Image sequence
SFT-Image source
Image prompt
Compute each expression and fill the correct result after '=' while leaving all other text unchanged. Output 1 image showing the final result filled in after each '='.
Video
Data video source
Video prompt
Compute each expression and fill the correct result after '=' while leaving all other text unchanged.
Circle Maximum Value
Image sequence
SFT-Image source
Image prompt
There are 5 numbers on the screen. Compare their numerical values, then draw exactly one red circle around the single number with the largest value. Do not circle any other number. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
There are 5 numbers on the screen. Compare their numerical values, then draw exactly one red circle around the single number with the largest value. Do not circle any other number. Show the circling process step by step. Only modify what the task requires; keep the background and any other element unchanged.
Connect Points A To C
Image sequence
SFT-Image source
Image prompt
Draw a straight red segment from point A to point C. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the figure with the completed segment AC drawn.
Video
Data video source
Video prompt
Draw a straight red segment from point A to point C and hold the completed connection. Only modify what the task requires; keep the background and any other element unchanged.
Glass Refraction
Image sequence
SFT-Image source
Image prompt
A horizontal air–glass interface is drawn with the glass region hatched below. A blue incident ray reaches the interface from air at θ = 58.3° from the normal (n_air = 1.00, n_glass = 1.995). Draw the red refracted ray inside the glass according to Snell's law, extending it to the image boundary. The angle annotation should be removed. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the incident and refracted rays.
Video
Data video source
Video prompt
A horizontal air–glass interface is drawn with the glass region hatched below. A blue incident ray reaches the interface from air at θ = 58.3° from the normal (n_air = 1.00, n_glass = 1.995). Draw the red refracted ray inside the glass according to Snell's law, extending it to the image boundary. The angle annotation should be removed. Only modify what the task requires; keep the background and any other element unchanged.
Klondike Card Reveal Progression
Image sequence
SFT-Image source
Image prompt
You are given a Klondike Solitaire board (green felt, stock on the left, four foundation piles on the right, seven tableau columns below). Only legal Klondike moves are allowed. Foundations build up by suit from Ace through King. Tableau builds down in rank alternating colors, and only a King may fill an empty tableau column. Make the single move that most directly advances the game priority: (1) move to a foundation if it follows suit-and-rank rules, else (2) make a tableau move that flips a hidden face-down card. If several moves tie, prefer the foundation. Show the resulting board after the move. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the board after the chosen move.
Video
Data video source
Video prompt
You are given a Klondike Solitaire board rendered above (green felt, stock on the left, four foundation piles on the right, seven tableau columns below). Only legal Klondike moves are allowed. Foundations build up by suit from Ace through King. Tableau builds down in rank alternating colors, and only a King may fill an empty tableau column.
From exactly this position, make the single move that most directly advances the game priority: (1) move to a foundation if it follows suit-and-rank rules, else (2) make a tableau move that flips a hidden face-down card. If several moves tie, prefer the foundation. Show the chosen card moving from its source pile to its destination, ending on the updated board. Only modify what the task requires; keep the background and any other element unchanged.
Line Intersection Point Identification
Image sequence
SFT-Image source
Image prompt
Extend the three line segments into full lines, then highlight in red the option at their common intersection point. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result with the extended lines and the correct option highlighted in red.
Video
Data video source
Video prompt
Extend the three line segments into full lines and keep the extensions visible; then highlight in red the option at their common intersection point.
Right Triangle Vertex Identification
Image sequence
SFT-Image source
Image prompt
Among the five markers, identify the triple that forms a right triangle and highlight in red only the marker at the 90-degree vertex. Output 1 image showing the final result with the right-angle vertex marker highlighted in red.
Video
Data video source
Video prompt
From five markers, locate the right-triangle triple and highlight only the 90-degree corner marker in red.
Hanoi
Image sequence
SFT-Image source
Image prompt
The image shows a Tower of Hanoi puzzle with 3 disks on 3 pegs (labelled 1–3 from left to right). Move all disks onto peg 3 so they end largest-on-bottom to smallest-on-top, following the rules: only the top disk of a peg may move, one disk at a time, and a larger disk may never rest on a smaller one. Output 5 images showing the tower after each move, in order. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows a Tower of Hanoi puzzle with 3 disks on 3 pegs (labelled 1–3 from left to right). Move all disks onto peg 3 so they end largest-on-bottom to smallest-on-top, following the rules: only the top disk of a peg may move, one disk at a time, and a larger disk may never rest on a smaller one. Generate a video that performs the optimal sequence of moves one at a time until every disk is stacked on peg 3. Only modify what the task requires; keep the background and any other element unchanged.
Multiple Occlusions Vertical
Image sequence
SFT-Image source
Image prompt
The scene shows 2 objects arranged in a horizontal line in the center of the frame, with a dark rectangular mask initially positioned above them. Move the mask vertically downward in a continuous motion until it leaves the frame. As it moves, the mask passes in front of the objects, temporarily blocking them from view. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the mask at uniform vertical progress from top to bottom.
Video
Data video source
Video prompt
The scene shows 2 objects arranged in a horizontal line in the center of the frame, with a dark rectangular mask initially positioned above them. Move the mask vertically downward in a continuous motion until it leaves the frame. As it moves, the mask passes in front of the objects, temporarily blocking them from view. Only modify what the task requires; keep the background and any other element unchanged.
Tic-Tac-Toe Forced Win Blocking Move
Image sequence
SFT-Image source
Image prompt
A Tic-Tac-Toe board with column labels A-C and row labels 1-3 shows the current X and O pieces and the player to move; the player's single best move (immediate win, winning setup, or forced block) has been played, with the played cell outlined to mark it. Output 1 image showing the final board after the move.
Video
Data video source
Video prompt
A Tic-Tac-Toe board is shown with column labels A-C and row labels 1-3, the current X and O pieces in place, and the player to move indicated at the top. Generate a video that plays the single best move for that player - the move that wins immediately, sets up a forced win, or blocks the opponent's threat - placing the new piece in its cell and outlining that cell to mark it. The camera is fixed; only the played move animates.
Identify One And Nine
Image sequence
SFT-Image source
Image prompt
The image shows a subset of digits chosen from 1 to 9 placed in different positions.
Find digit 1 and digit 9.
Only circle digits '1' and '9'. Do not circle other digits.
Draw a red circle around each target digit. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The image shows a subset of digits chosen from 1 to 9 placed in different positions.
Find digit 1 and digit 9.
Only circle digits '1' and '9'. Do not circle other digits.
Draw a red circle around each target digit. Only modify what the task requires; keep the background and any other element unchanged.
Multiple Bounces
Image sequence
SFT-Image source
Image prompt
A black ball starts with the shown initial direction arrow. It moves in straight lines and reflects off the two vertical walls following the law of reflection (the angle of reflection equals the angle of incidence). Follow this unique trajectory until the ball hits one of the red hollow circle targets in the lower-right area. Identify which target circle is hit and show the full trajectory, drawn as an orange line. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the ball's final position at the hit target, with the complete trajectory drawn as an orange line.
Video
Data video source
Video prompt
A black ball starts with the shown initial direction arrow. It moves in straight lines and reflects off the two vertical walls following the law of reflection (the angle of reflection equals the angle of incidence). Follow this unique trajectory until the ball hits one of the red hollow circle targets in the lower-right area. Identify which target circle is hit and show the full trajectory, drawn as an orange line, step by step. Only modify what the task requires; keep the background and any other element unchanged.
Gravity Physics
Image sequence
Data image source
Image prompt
A ball at height 12.9m with initial downward velocity 2.0 m/s falls under gravity 7.1 m/s² and bounces on ground with elasticity 0.70. Only modify what the task requires; keep the background and any other element unchanged. Output one image at each endpoint of the trajectory — at the top (peak) of every bounce and at the bottom each time the ball hits the ground and rebounds — each with a velocity arrow showing direction and magnitude, until the ball comes to rest.
Video
Data video source
Video prompt
A ball at height 12.9m with initial downward velocity 2.0 m/s falls under gravity 7.1 m/s² and bounces on ground with elasticity 0.70. Show the full trajectory with velocity arrows (direction and magnitude) updating throughout until the ball stops. Only modify what the task requires; keep the background and any other element unchanged.
Dot To Dot Outline Drawing
Image sequence
SFT-Image source
Image prompt
Connect numbered dots from 1 to 25 in exact order with a continuous black line, keeping all dot labels visible. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the completed outline with all dots connected.
Video
Data video source
Video prompt
Connect numbered dots from 1 to 25 in exact order with a continuous black line, keeping all dot labels visible. Only modify what the task requires; keep the background and any other element unchanged.
Parallel Line Through Circle
Image sequence
SFT-Image source
Image prompt
Identify the given line segment; draw a new line through the small circle that is parallel to that segment and keep it visible; and highlight in red the option lying on the new line. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final figure with the parallel line drawn and the correct option marked red.
Video
Data video source
Video prompt
Identify the existing line segment first, draw a parallel line through the small circle, keep it visible, and mark in red the option that lies on the constructed line. Only modify what the task requires; keep the background and any other element unchanged.
Point Reflection Across Line
Image sequence
SFT-Image source
Image prompt
Project the point perpendicularly onto the mirror line, then continue the same distance across the line to construct its reflection, keeping the full construction visible. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final construction with the reflected point and the projection segments drawn.
Video
Data video source
Video prompt
Use projection and equal-distance mirroring across the mirror line to build the reflected point, keeping both dashed segments and the mirror line visible.
Locate Twelve O Clock Arrows
Image sequence
SFT-Image source
Image prompt
The image contains 4 clocks, each with only an hour hand. Exactly one clock has its hour hand pointing to 12 o'clock. First find the single clock pointing to 12 o'clock, then draw a red circle around it. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The image contains 4 clocks, each with only an hour hand. Exactly one clock has its hour hand pointing to 12 o'clock. First find the single clock pointing to 12 o'clock, then draw a red circle around it. Only modify what the task requires; keep the background and any other element unchanged. Show the complete solution step by step.
Color Mixing
Image sequence
SFT-Image source
Image prompt
The scene shows two solid-colored circular light sources on the left and right of a black canvas, and a white-bordered square mixing zone in the centre. In additive (RGB) color mixing, when two lights overlap their channels add together and each channel is clamped at 255: result_R = min(R1 + R2, 255); result_G = min(G1 + G2, 255); result_B = min(B1 + B2, 255). The left light is an RGB(104, 0, 85) colored light and the right light is an RGB(37, 8, 88) colored light. Compute the additive mix channel by channel and fill the white-bordered mixing zone with the resulting color, keeping everything else unchanged. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the mixing zone filled with the mixed color.
Video
Data video source
Video prompt
The scene shows two solid-colored circular light sources on the left and right of a black canvas, and a white-bordered square mixing zone in the centre. In additive (RGB) color mixing, when two lights overlap their channels add together and each channel is clamped at 255: result_R = min(R1 + R2, 255); result_G = min(G1 + G2, 255); result_B = min(B1 + B2, 255). The left light is an RGB(104, 0, 85) colored light and the right light is an RGB(37, 8, 88) colored light. Compute the additive mix channel by channel and fill the white-bordered mixing zone with the resulting color, keeping everything else unchanged. Only modify what the task requires; keep the background and any other element unchanged.
Ball Bounces Given Time
Image sequence
SFT-Image source
Image prompt
A ball is placed at the initial position with a direction arrow indicating its movement direction. Simulate the ball bouncing 5 times off the boundary walls following elastic collision physics (angle of incidence equals angle of reflection). The ball stops after the 5th bounce, with its final position at the wall where the last collision occurs. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the ball's final position, with the complete trajectory path drawn as an orange line showing the ball's route including all bounce points.
Video
Data video source
Video prompt
A ball is placed at the initial position with a direction arrow indicating its movement direction. Simulate the ball bouncing 5 times off the boundary walls following elastic collision physics (angle of incidence equals angle of reflection). The ball stops after the 5th bounce, with its final position at the wall where the last collision occurs. Only modify what the task requires; keep the background and any other element unchanged.
Snake Eat Food Path
Image sequence
SFT-Image source
Image prompt
Show the snake movement without self-collision; after eating the red food, the snake length increases by exactly one cell. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image of the board right after the snake eats the food (the snake is now one cell longer). On top of that board, draw the head's full travel path from its starting cell to the food as a dotted orange line through the cell centres, with a small open circle on the start cell and a solid arrowhead pointing into the food cell.
Video
Data video source
Video prompt
Show the snake movement process without self-collision; after eating the red food, the snake length increases by exactly one cell. Only modify what the task requires; keep the background and any other element unchanged.
Xiangqi Mate In One
Image sequence
SFT-Image source
Image prompt
Xiangqi rules: red and black alternate one legal move each turn; a side wins by checkmating the opponent's general. It is red to move. Make one legal move that delivers immediate checkmate. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board after making the checkmating move.
Video
Data video source
Video prompt
Xiangqi rules: red and black alternate one legal move each turn; a side wins by checkmating the opponent's general. It is red to move. Make one legal move that delivers immediate checkmate. Only modify what the task requires; keep the background and any other element unchanged.
Triangle Orthocenter Identification
Image sequence
SFT-Image source
Image prompt
Find the orthocenter of the triangle (the intersection point of the triangle's altitudes) and mark the correct labeled option in red. Output 1 image showing the final figure with the correct option marked red.
Video
Data video source
Video prompt
Find the orthocenter of the triangle and mark the correct option in red.
Perpendicular Line Through Circle
Image sequence
SFT-Image source
Image prompt
Draw a line through the small circle that is perpendicular to the given line, then highlight in red the option lying on that perpendicular line. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result with the perpendicular line drawn and the correct option highlighted in red.
Video
Data video source
Video prompt
From the given line direction, draw the perpendicular through the small circle, keep the new line on screen, and highlight in red the option on that line. Only modify what the task requires; keep the background and any other element unchanged.
Identify Chinese Character
Image sequence
Data image source
Image prompt
Find and circle the Chinese character among the displayed characters. Only one character is Chinese. Draw a red circle around it. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
Find and circle the Chinese character among the displayed characters. Only one character is Chinese. Draw a red circle around it. Only modify what the task requires; keep the background and any other element unchanged.
Read The Chart Data Semantic Comprehension
Image sequence
SFT-Image source
Image prompt
The scene shows a expense table with Department as rows and Product as columns. Find the minimum value within the column corresponding to the Product 'Tablet' and draw a red rectangular border around the corresponding cell to highlight it. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
The scene shows a expense table with Department as rows and Product as columns. Find the minimum value within the column corresponding to the Product 'Tablet' and draw a red rectangular border around the corresponding cell to highlight it. Only modify what the task requires; keep the background and any other element unchanged.
Chess Mate In One
Image sequence
SFT-Image source
Image prompt
White to move. Play the unique move that gives checkmate in one. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board after making the checkmating move.
Video
Data video source
Video prompt
White to move. Play the unique move that gives checkmate in one. Show the move being played, ending on the checkmate position. Only modify what the task requires; keep the background and any other element unchanged.
Sudoku Stepwise Completion
Image sequence
SFT-Image source
Image prompt
Solve the 4x4 Sudoku puzzle. Following Sudoku rules for rows, columns, and 2x2 subgrids, fill in the correct missing number in every blank cell. Do not change any other content. Output 1 image showing the fully solved grid.
Video
Data video source
Video prompt
Solve the 4x4 Sudoku puzzle visually. Following Sudoku rules, move the cursor cell by cell and fill only the blank cells step by step until the grid is complete. Do not change any other content.
Dot To Dot Task
Image sequence
Data image source
Image prompt
The scene shows 8 numbered dots scattered across the image. Connect the dots in numerical order (1→2→3→...→8) by drawing red straight lines between them, one line at a time in sequence. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each new line connection, showing all connections made so far.
Video
Data video source
Video prompt
The scene shows 8 numbered dots scattered across the image. Connect the dots in numerical order (1→2→3→...→8) by drawing red straight lines between them, one line at a time in sequence. Only modify what the task requires; keep the background and any other element unchanged.
Triangle Center Identification
Image sequence
SFT-Image source
Image prompt
Among five markers, three form a triangle and one marker is its centroid; highlight only the centroid marker in red. Output 1 image showing the final result with the centroid marker highlighted in red.
Video
Data video source
Video prompt
Among five markers, three form a triangle and one marker is its centroid; highlight only the centroid marker in red.
Grid Highest Cost
Image sequence
SFT-Image source
Image prompt
A 4x4 grid shows integer costs in each cell. The top-left cell is green (start) and the bottom-right cell is red (goal). A yellow Pac-Man starts on the green cell. Find a path from start to goal that maximizes the sum of costs collected (each time Pac-Man enters a cell, add that cell's cost). Only orthogonal moves (up/down/left/right). Keep the grid lines and cell cost numbers unchanged; only Pac-Man should move. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the complete highest-cost path drawn as an orange line from the green start cell to the red goal cell.
Video
Data video source
Video prompt
A 4x4 grid shows integer costs in each cell. The top-left cell is green (start) and the bottom-right cell is red (goal). A yellow Pac-Man starts on the green cell. Animate Pac-Man moving step-by-step along a path from start to goal that maximizes the sum of costs collected (each time Pac-Man enters a cell, add that cell's cost). Only orthogonal moves (up/down/left/right). Keep the grid lines and cell cost numbers unchanged; only Pac-Man should move. Only modify what the task requires; keep the background and any other element unchanged.
Tic-Tac-Toe Next Move O
Image sequence
SFT-Image source
Image prompt
Tic-Tac-Toe rules: players alternate placing one mark in an empty cell on a 3x3 grid, and three in a row (row/column/diagonal) wins. It is 'O' to move. Place one 'O' in the best empty cell: win if possible, otherwise block 'X', otherwise play the strongest move. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board after placing the 'O'.
Video
Data video source
Video prompt
Tic-Tac-Toe rules: players alternate placing one mark in an empty cell on a 3x3 grid, and three in a row (row/column/diagonal) wins. It is 'O' to move. Place one 'O' in the best empty cell: win if possible, otherwise block 'X', otherwise play the strongest move. Only modify what the task requires; keep the background and any other element unchanged.
Light Reflection Ray Identification
Image sequence
SFT-Image source
Image prompt
Draw the reflected light ray from the small circle off the mirror, then mark the correct option in red. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result with the reflected ray drawn and the correct option highlighted in red.
Video
Data video source
Video prompt
Trace the reflection from the small circle to the mirror and choose the matching option.
Freecell Valid Card Transfer
Image sequence
SFT-Image source
Image prompt
Make the single most helpful legal move on this FreeCell board (eight tableau columns, four free cells, four foundations). Tableau builds downward with alternating red/black, one card moves per step, and foundations build up by suit from the ace. Show the resulting board after the move. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the board after the best transfer.
Video
Data video source
Video prompt
Generate a video that makes the single most helpful legal move on this FreeCell board (eight tableau columns, four free cells, four foundations). Tableau builds downward with alternating red/black, one card moves per step, and foundations build up by suit from the ace. Show the chosen card moving from its source to its destination, ending on the updated board. Only modify what the task requires; keep the background and any other element unchanged.
Fluid Diffusion Reasoning
Image sequence
SFT-Image source
Image prompt
An ink droplet falls from above the center of a glass beaker filled with water. Upon entering the water, the ink forms irregular downward-extending tendrils due to gravity and initial impact. The ink then diffuses through the water, creating swirling patterns and eddies, until it reaches a stable state of uniform color distribution throughout the entire volume of water. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
An ink droplet falls from above the center of a glass beaker filled with water. Upon entering the water, the ink forms irregular downward-extending tendrils due to gravity and initial impact. The ink then diffuses through the water, creating swirling patterns and eddies, until it reaches a stable state of uniform color distribution throughout the entire volume of water. Only modify what the task requires; keep the background and any other element unchanged.
Minesweeper Guaranteed Mine Cells
Image sequence
SFT-Image source
Image prompt
Apply Minesweeper deduction from the revealed numbers: combining all the clues together, flag every cell that is logically certain to be a mine. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the board with all guaranteed mine cells flagged.
Video
Data video source
Video prompt
Apply Minesweeper deduction from the revealed numbers: combining all the clues together, flag every cell that is logically certain to be a mine. Animate these guaranteed mine flags step by step. Only modify what the task requires; keep the background and any other element unchanged.
Communicating Vessels
Image sequence
Data image source
Image prompt
A system of 4 communicating vessels with equal-diameter vertical tubes is filled with honey (honey-like (high viscosity)), which appears orange in color. As shown in the initial frame, the liquid levels in the tubes are [23, 22, 55, 12] cm respectively. Due to pressure differences between the tubes, the liquid begins to flow through the connecting channels at the bottom. The flow is governed by hydrostatic pressure equalization and damped by viscous resistance with coefficient k=1.87. As the liquid redistributes, the height differences gradually decrease, and the system evolves toward equilibrium. Eventually, through conservation of volume, all tubes reach the same final liquid level, which equals the average of the initial heights. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the liquid levels at uniform time progress during equalization.
Video
Data video source
Video prompt
A system of 4 communicating vessels with equal-diameter vertical tubes is filled with honey (honey-like (high viscosity)), which appears orange in color. As shown in the initial frame, the liquid levels in the tubes are [23, 22, 55, 12] cm respectively. Due to pressure differences between the tubes, the liquid begins to flow through the connecting channels at the bottom. The flow is governed by hydrostatic pressure equalization and damped by viscous resistance with coefficient k=1.87. As the liquid redistributes, the height differences gradually decrease, and the system evolves toward equilibrium. Eventually, through conservation of volume, all tubes reach the same final liquid level, which equals the average of the initial heights. Simulate this settling process from the initial unbalanced state to the final stable equilibrium. Only modify what the task requires; keep the background and any other element unchanged.
Domino Chain Branch Path Prediction
Image sequence
SFT-Image source
Image prompt
Push the green START domino to begin the chain reaction. Depict the complete domino falling process across the image sequence. The dominos fall in order from left to right. The gray trunk falls first, then splits into the red Branch A (upward) and blue Branch B (downward). Both branches complete successfully. The final state shows all dominos fallen and tilted right. Only modify what the task requires; keep the background and any other element unchanged. Output one image each time a domino finishes falling: one image per trunk-domino fall, then one image per branch step (A and B fall in lock-step). The last image shows the final state.
Video
Data video source
Video prompt
Push the green START domino to begin the chain reaction. Show the complete domino falling process. The dominos fall step by step from left to right. The gray trunk falls first, then splits into the red Branch A (upward) and blue Branch B (downward). Branch A has a gap marked with a red X that stops the chain reaction in that branch. The final state shows Branch A with fallen dominos before the gap and standing dominos after, while Branch B completes fully. Only modify what the task requires; keep the background and any other element unchanged.
Tic-Tac-Toe Next Move X
Image sequence
SFT-Image source
Image prompt
Tic-Tac-Toe rules: players alternate placing one mark in an empty cell on a 3x3 grid, and three in a row (row/column/diagonal) wins. It is 'X' to move. Place one 'X' in the best empty cell: win if possible, otherwise block 'O', otherwise play the strongest move. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board after placing the 'X'.
Video
Data video source
Video prompt
Tic-Tac-Toe rules: players alternate placing one mark in an empty cell on a 3x3 grid, and three in a row (row/column/diagonal) wins. It is 'X' to move. Place one 'X' in the best empty cell: win if possible, otherwise block 'O', otherwise play the strongest move. Only modify what the task requires; keep the background and any other element unchanged.
Faucet Fill Correct Containers
Image sequence
SFT-Image source
Image prompt
Open the faucet so water flows through the connected pipes into the reachable cups, ending with water filling the connected containers; cups outside the connected network stay unfilled. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final filled state.
Video
Data video source
Video prompt
Open the faucet, animate water flowing through connected pipes into reachable cups, and show water filling the connected containers; cups outside the connected network stay unfilled.
Clock
Image sequence
SFT-Image source
Image prompt
The clock shows 3:09. Show what the clock will look like after 7 hours. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images at uniform time progress (1/4, 2/4, ..., 4/4) of the elapsed 7-hour interval.
Video
Data video source
Video prompt
The clock shows 3:09. Show what the clock will look like after 7 hours. Only modify what the task requires; keep the background and any other element unchanged.
Hit Target After Bounce
Image sequence
SFT-Image source
Image prompt
The scene shows a ball with an arrow indicating its initial direction, and several empty target positions (hollow circles) on the right side. Simulate the ball moving along this direction and bouncing off walls following the law of reflection (the angle of reflection equals the angle of incidence). The ball will follow a complete trajectory and eventually align exactly with and completely overlap one of the target positions. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the ball's final position, with the complete trajectory path drawn as a yellow line including all bounce points.
Video
Data video source
Video prompt
The scene shows a ball with an arrow indicating its initial direction, and several empty target positions (hollow circles) on the right side. Simulate the ball moving along this direction and bouncing off walls following the law of reflection (the angle of reflection equals the angle of incidence). The ball will follow a complete trajectory and eventually align exactly with and completely overlap one of the target positions. Only modify what the task requires; keep the background and any other element unchanged.
Block Slide Final Rest Position
Image sequence
SFT-Image source
Image prompt
An inclined plane with a frictional sliding block is shown. Given θ, μ_slope, μ_plane, and L_slope, predict the block’s final rest position. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the block sliding to its final rest position at uniform stages.
Video
Data video source
Video prompt
An inclined plane with a frictional sliding block is shown. Given θ, μ_slope, μ_plane, and L_slope, predict the block’s final rest position. The final frame should display where the block stops.
Binairo
Image sequence
SFT-Image source
Image prompt
The image shows a Binairo (Takuzu) grid: fill every cell with 0 or 1. Each row and column has the same count of 0s and 1s. No three adjacent cells in a row or column may be the same. No two rows may be identical; no two columns may be identical. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows a Binairo (Takuzu) grid: fill every cell with 0 or 1. Each row and column has the same count of 0s and 1s. No three adjacent cells in a row or column may be the same. No two rows may be identical; no two columns may be identical. Generate a video that completes the puzzle by revealing the missing digits step by step. Only modify what the task requires; keep the background and any other element unchanged.
Life Game Next Generation Prediction
Image sequence
SFT-Image source
Image prompt
Conway's Game of Life on a grid of cells, each alive (black) or dead (white). For every cell, count its 8 neighbours (the 4 orthogonal + 4 diagonal cells); cells outside the grid count as dead. Then update all cells at the same time: a live cell with 2 or 3 live neighbours stays alive, otherwise it dies; a dead cell with exactly 3 live neighbours becomes alive. The number written in a cell is its live-neighbour count; in a result board, a green outline marks a cell newly born that step and a red outline marks a cell that just died. Simulate the number of generations shown in the image header ("Generations: N"). Determine the next-generation board. For each generation, output two images: first the live-neighbour count written in every cell, then the board after that generation updates (cells newly born outlined green, cells that died outlined red). Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Conway's Game of Life on a grid of cells, each alive (black) or dead (white). For every cell, count its 8 neighbours (the 4 orthogonal + 4 diagonal cells); cells outside the grid count as dead. Then update all cells at the same time: a live cell with 2 or 3 live neighbours stays alive, otherwise it dies; a dead cell with exactly 3 live neighbours becomes alive. The number written in a cell is its live-neighbour count; in a result board, a green outline marks a cell newly born that step and a red outline marks a cell that just died. Simulate the number of generations shown in the image header ("Generations: N"). Show the live-neighbour count in every cell, then reveal the next-generation board. Only modify what the task requires; keep the background and any other element unchanged.
Sequence Completion
Image sequence
SFT-Image source
Image prompt
The scene shows a power sequence. Elements are arranged horizontally from left to right. The last position contains a question mark (?) indicating a missing element. Observe the pattern: each number is the square of consecutive integers (1², 2², 3², etc.). Determine the element that should replace the question mark to complete the sequence according to the established pattern. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
The scene shows a power sequence. Elements are arranged horizontally from left to right. The last position contains a question mark (?) indicating a missing element. Observe the pattern: each number is the square of consecutive integers (1², 2², 3², etc.). Determine the element that should replace the question mark to complete the sequence according to the established pattern. Only modify what the task requires; keep the background and any other element unchanged.
Symbol Substitute
Image sequence
Data image source
Image prompt
Substitute the solid heart at position 2 with a orange solid left triangle. A reference panel in the top-right shows the target symbol. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result, in which the old symbol has been removed and the new symbol appears at the same position in its full color, with every other position left unchanged.
Video
Data video source
Video prompt
Substitute the solid heart at position 2 with a orange solid left triangle. A reference panel in the top-right shows the target symbol. The animation shows the old symbol smoothly fading out into the white background, then the new symbol smoothly fading in from white to its full color at the same position. Only modify what the task requires; keep the background and any other element unchanged.
Hexagon Pattern Completion
Image sequence
SFT-Image source
Image prompt
Six triangles meet at the center inside a hexagon outline. Five show solid colors; one shows '?'. The visible colors follow a repeating ABCABC pattern clockwise. Choose the color that continues that pattern in the missing slot, paint it in, and remove the question mark without altering anything else. Output 1 image showing the final hexagon with the missing wedge filled.
Video
Data video source
Video prompt
Six triangles meet at the center inside a hexagon outline. Five show solid colors; one shows '?'. The visible colors follow a repeating ABCABC pattern clockwise. Choose the color that continues that pattern in the missing slot, paint it in, and remove the question mark without altering anything else.
Shape Scaling
Image sequence
Data image source
Image prompt
The image shows an analogy A:B :: C:?. A and C are shapes in their original sizes, B is the scaled version of A, and ? is the missing scaled version of C. In the example A → B, the diamond is rescaled by a fixed factor. Infer this scaling factor by comparing A and B, then apply the same factor to C (the square) to decide what ? should look like as the correctly scaled version of C. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the completed analogy.
Video
Data video source
Video prompt
The image shows an analogy A:B :: C:?. A and C are shapes in their original sizes, B is the scaled version of A, and ? is the missing scaled version of C. In the example A → B, the diamond is rescaled by a fixed factor. Infer this scaling factor by comparing A and B, then apply the same factor to C (the square) to decide what ? should look like as the correctly scaled version of C. Only modify what the task requires; keep the background and any other element unchanged.
Wood Slide
Image sequence
SFT-Image source
Image prompt
Shown is a wooden sliding-block puzzle. Each piece can be pushed up, down, left, or right into adjacent empty squares, but never lifted out, turned, or overlapped with another piece. Maneuver the marked target block to the dashed goal cell. Output 1 image showing the board after each move, in order. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Shown is a wooden sliding-block puzzle. Each piece can be pushed up, down, left, or right into adjacent empty squares, but never lifted out, turned, or overlapped with another piece. Maneuver the marked target block to the dashed goal cell. Generate a video that animates a sequence of slides, moving one block at a time, and ends with the target seated on the goal. Only modify what the task requires; keep the background and any other element unchanged.
Ballcolor
Image sequence
SFT-Image source
Image prompt
In the scene, there are 4 ball clusters: red cluster A has 5 balls, and there are other clusters with different numbers of balls. The global TOTAL label shows the sum 25. Control the red cluster A to absorb other clusters. You can only absorb clusters that are smaller than the red cluster. Find the correct sequence to absorb all clusters step by step until all balls become red. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each merge event, showing the state after the red cluster absorbs a smaller cluster.
Video
Data video source
Video prompt
In the scene, there are 4 ball clusters: red cluster A has 5 balls, and there are other clusters with different numbers of balls. The global TOTAL label shows the sum 25. Control the red cluster A to absorb other clusters. You can only absorb clusters that are smaller than the red cluster. Find the correct sequence to absorb all clusters step by step until all balls become red. Only modify what the task requires; keep the background and any other element unchanged.
Eulero
Image sequence
SFT-Image source
Image prompt
Solve the Eulero puzzle in the image: complete the grid with letter-number pairs so that letters form a Latin square, numbers form a Latin square, and no pair repeats anywhere in the grid. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Solve the Eulero puzzle in the image: complete the grid with letter-number pairs so that letters form a Latin square, numbers form a Latin square, and no pair repeats anywhere in the grid. Generate a video that reveals the correct pairs cell by cell until the whole grid is filled. Only modify what the task requires; keep the background and any other element unchanged.
Pac-Man Ghost Avoidance Next Move
Image sequence
SFT-Image source
Image prompt
Observe the current Pac-Man maze state. Identify Pac-Man, the ghosts, walls, and walkable corridors, then decide the safest immediate next move. The answer should avoid immediate danger and preserve future escape options. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the board after Pac-Man makes the safest next move.
Video
Data video source
Video prompt
Given the current Pac-Man maze state, determine the safest immediate next move, rank the candidate legal moves, and explain which choice best preserves future escape options against the nearby ghosts. Only modify what the task requires; keep the background and any other element unchanged.
Symbol Delete
Image sequence
Data image source
Image prompt
Delete the red solid diamond (suit) at position 7. In the result the target symbol has faded to the background colour until it is no longer visible, then the remaining symbols have slid leftward to close the gap and form the post-deletion arrangement. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final post-deletion result.
Video
Data video source
Video prompt
Delete the red solid diamond (suit) at position 7. The animation shows the target symbol smoothly fading to the background colour until it is no longer visible, then the remaining symbols slide leftward to close the gap and form the post-deletion arrangement. Only modify what the task requires; keep the background and any other element unchanged.
Campsite
Image sequence
SFT-Image source
Image prompt
The image shows a Campsite (Tents) puzzle: a grid containing trees, with a number to the right of each row and below each column. Place tents on empty cells so that (1) each tent is orthogonally adjacent to at least one tree, (2) no two tents are adjacent even diagonally, and (3) the number of tents in each row and column matches the given counts. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows a Campsite (Tents) puzzle: a grid containing trees, with a number to the right of each row and below each column. Place tents on empty cells so that (1) each tent is orthogonally adjacent to at least one tree, (2) no two tents are adjacent even diagonally, and (3) the number of tents in each row and column matches the given counts. Generate a video that places the tents one at a time until the puzzle is fully solved. Only modify what the task requires; keep the background and any other element unchanged.
Skyscraper
Image sequence
SFT-Image source
Image prompt
The image shows a Skyscraper puzzle: an N×N grid whose cells hold building heights 1 to N, each height appearing exactly once in every row and every column. The number outside a row or column is a clue — standing at that edge and looking inward along the line, it tells how many buildings are visible, where a taller building completely hides every shorter building behind it, so a building is seen only if no taller building stands between it and that edge. Some cells are already filled. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows a Skyscraper puzzle: an N×N grid whose cells hold building heights 1 to N, each height appearing exactly once in every row and every column. The number outside a row or column is a clue — standing at that edge and looking inward along the line, it tells how many buildings are visible, where a taller building completely hides every shorter building behind it, so a building is seen only if no taller building stands between it and that edge. Some cells are already filled. Generate a video that fills in the grid until it is fully and correctly solved. Only modify what the task requires; keep the background and any other element unchanged.
Construction Stack
Image sequence
Data image source
Image prompt
Match the left CURRENT stacks to the TARGET on the right by moving blocks (one top block per move). Keep the TARGET stacks on the right unchanged. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each block move, showing the updated stack state.
Video
Data video source
Video prompt
Match the left CURRENT stacks to the TARGET on the right by moving blocks (one top block per move). Keep the TARGET stacks on the right unchanged. Only modify what the task requires; keep the background and any other element unchanged.
Map
Image sequence
SFT-Image source
Image prompt
Shown is a four-coloring puzzle: a map split into labeled regions, some pre-colored red, green, blue, or yellow. Complete the coloring of the gray regions with those four colors so that every pair of regions that shares an edge has different colors. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Shown is a four-coloring puzzle: a map split into labeled regions, some pre-colored red, green, blue, or yellow. Complete the coloring of the gray regions with those four colors so that every pair of regions that shares an edge has different colors. Generate a video that colors each remaining region in turn until the whole map is colored. Only modify what the task requires; keep the background and any other element unchanged.
N-Queens
Image sequence
SFT-Image source
Image prompt
You are given an 8×8 chessboard with a few queens already in place. Complete the 8-Queens puzzle by placing the remaining queens so that no queen can capture another along any row, column, or diagonal. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
You are given an 8×8 chessboard with a few queens already in place. Complete the 8-Queens puzzle by placing the remaining queens so that no queen can capture another along any row, column, or diagonal. Generate a video that fills in the missing queens step by step, finishing on the full solution. Only modify what the task requires; keep the background and any other element unchanged.
Nonogram
Image sequence
SFT-Image source
Image prompt
Picross/Nonogram: the numbers beside each row and above each column give the lengths of consecutive shaded blocks (in order). Blocks must be separated by at least one blank. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Picross/Nonogram: the numbers beside each row and above each column give the lengths of consecutive shaded blocks (in order). Blocks must be separated by at least one blank. Generate a video that fills in the solution cell by cell until the completed picture is shown. Only modify what the task requires; keep the background and any other element unchanged.
Shape Scale Then Outline
Image sequence
Data image source
Image prompt
Apply the two-step transformation: scale change followed by fill-to-outline conversion. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the completed analogy.
Video
Data video source
Video prompt
Animate the two-step transformation: scale change followed by fill-to-outline conversion. Only modify what the task requires; keep the background and any other element unchanged.
Bookshelf
Image sequence
SFT-Image source
Image prompt
In the scene, there is a bookshelf with a set of books already placed, and a few purple books waiting on the right. There are gaps between the books on the shelf. Place each purple book into a gap where its height is closest to the surrounding books. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each book is inserted, showing the updated shelf.
Video
Data video source
Video prompt
In the scene, there is a bookshelf with a set of books already placed, and a few purple books waiting on the right. There are gaps between the books on the shelf. Place each purple book into a gap where its height is closest to the surrounding books. Show the insertion process step by step. Only modify what the task requires; keep the background and any other element unchanged.
2D Pattern Completion
Image sequence
SFT-Image source
Image prompt
A honeycomb of shaded hexagonal cells forms a larger motif. Several cells are left empty while the rest expose a strict rule—either a linear color progression across the lattice or perfect bilateral symmetry. Infer the rule and restore every blank so rows and columns stay internally consistent. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the fully completed, coherent pattern.
Video
Data video source
Video prompt
A fixed orthographic view shows a honeycomb of shaded hexagonal cells forming a larger motif. Several cells are left empty while the rest expose a strict rule—either a linear color progression across the lattice or perfect bilateral symmetry. Infer the rule and mentally restore every blank. The ground-truth clip should progressively paint the missing hexes so rows and columns stay internally consistent, ending with a fully completed, coherent pattern. Only modify what the task requires; keep the background and any other element unchanged.
Colored Bar Order Matching
Image sequence
SFT-Image source
Image prompt
What is the color of the missing part denoted with a question mark? This part should be completely filled with the correct color while the other parts should be unchanged. The question mark is removed. Output 1 image showing the final bars with the missing bar filled in the correct color.
Video
Data video source
Video prompt
Infer the missing color by matching bar heights: equal-height bars have equal colors. Complete the '?' bar with the correct color only, do not alter any other bar, and keep the camera and layout fixed after the mark disappears.
Chess Ranger Attack Zone Reasoning
Image sequence
SFT-Image source
Image prompt
An 8x8 chessboard shows a position with standard chess pieces for both sides, each attacking along its normal chess lines. One piece is outlined: it is the target piece and belongs to the side to move, which is named above the board. Work out the squares the enemy pieces attack, the safe squares the target piece can move to (squares the enemy does not attack), and its single best safe move. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image of the resulting board: keep the target piece outlined, mark its safe destination squares, and draw an arrow from the target piece to its best move's square.
Video
Data video source
Video prompt
An 8x8 chessboard holds a position with standard pieces on both sides. The single outlined piece belongs to the side to move and is the one to relocate. Generate a video that builds the answer in stages: the enemy's attacked region, the target piece's safe destination squares, the candidate moves, and finally an arrow showing its best safe move.
Futoshiki
Image sequence
SFT-Image source
Image prompt
The image shows a Futoshiki puzzle: an N×N grid to fill with the numbers 1 to N so that each number appears exactly once in every row and every column, and every inequality sign (>, <, ^, v) between two adjacent cells is respected. Some cells are pre-filled. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows a Futoshiki puzzle: an N×N grid to fill with the numbers 1 to N so that each number appears exactly once in every row and every column, and every inequality sign (>, <, ^, v) between two adjacent cells is respected. Some cells are pre-filled. Generate a video that fills in the remaining cells one at a time until the whole grid is solved. Only modify what the task requires; keep the background and any other element unchanged.
Rotation Puzzle
Image sequence
SFT-Image source
Image prompt
Solve this pipe puzzle. The scene shows four square tiles, each containing a pipe segment. The tiles stay fixed in place and do not rotate; only the pipe inside each tile rotates. Rotate the pipe within each tile so that all pipe segments link up into a single continuous path. All pipes rotate at the same time and reach their final orientations together. Keep the camera view fixed in the top-down perspective and keep every tile in its original position. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images at uniform intervals showing the rotation progress, ending with the solved puzzle.
Video
Data video source
Video prompt
Solve this pipe puzzle. The scene shows four square tiles, each containing a pipe segment. The tiles stay fixed in place and do not rotate; only the pipe inside each tile rotates. Rotate the pipe within each tile so that all pipe segments link up into a single continuous path. All pipes rotate at the same time and reach their final orientations together. Keep the camera view fixed in the top-down perspective and keep every tile in its original position. Stop the video when all pipes are connected and the puzzle is solved. Only modify what the task requires; keep the background and any other element unchanged.
Circle Cluster Pattern Completion
Image sequence
SFT-Image source
Image prompt
A cluster of shapes is arranged along a circular arc. Each shape belongs to a size category based on its size. Use the rule: shapes with the same size category must have the same color. One shape is missing its fill color and is marked with a '?'. Infer the correct color by matching its size to other shapes of the same size category, then fully fill the missing shape with that solid color. The '?' must disappear. Do not move, resize, or recolor any other shape. Output 1 image showing the final scene with the missing shape filled.
Video
Data video source
Video prompt
A cluster of shapes is arranged along a circular arc. Each shape belongs to a size category based on its size. Use the rule: shapes with the same size category must have the same color. One shape is missing its fill color and is marked with a '?'. Infer the correct color by matching its size to other shapes of the same size category, then fully fill the missing shape with that solid color. The '?' must disappear. Do not move, resize, or recolor any other shape. After filling, the scene remains perfectly still.
Rhythm Game Note Hit Sequence
Image sequence
SFT-Image source
Image prompt
A multi-lane rhythm-game chart shows coloured notes; time runs from top to bottom and notes on the same row are hit together. The timing_order list on the right gives, line by line as n. followed by the lanes, the lanes hit at each row from top to bottom. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final chart with the complete timing_order list.
Video
Data video source
Video prompt
A multi-lane rhythm-game chart shows coloured notes; time runs from top to bottom and notes on the same row are hit at the same time. Generate a video that scans downward row by row: each time the scan reaches a row of notes, hit them and add the next line, numbered n. followed by the list of lanes hit at that row, to the timing_order list on the right. Build the list up one line at a time, top to bottom, ending with the full list. The camera stays fixed.
Control Panel
Image sequence
Data image source
Image prompt
The image shows a control panel with three identical control units. Each unit has a colored indicator light at the top and a control lever at the bottom that can be moved to three positions (left, middle, or right). Each control unit has its own single indicator light, whose color is determined solely by that unit's lever position.
Observe the current control panel to infer the relationship between lever positions and light colors. Based on this inferred relationship, adjust the levers that need to be changed to make all indicator lights show cyan color. Only modify what the task requires; keep the background and any other element unchanged. Process the levers from left to right and output one image after each single lever move.
Video
Data video source
Video prompt
The image shows a control panel with three identical control units. Each unit has a colored indicator light at the top and a control lever at the bottom that can be moved to three positions (left, middle, or right). Each control unit has its own single indicator light, whose color is determined solely by that unit's lever position.
Observe the current control panel to infer the relationship between lever positions and light colors. Based on this inferred relationship, adjust the levers that need to be changed to make all indicator lights show cyan color. Only modify what the task requires; keep the background and any other element unchanged.
Thermometers
Image sequence
SFT-Image source
Image prompt
You are given a grid fully covered by thermometers — straight segments of two or more cells, each with a circular bulb marking one end. Mercury always flows from the bulb and stays contiguous, so each thermometer is filled as a run starting at its bulb. The numbers outside the grid (one per row on the right, one per column on the bottom) state how many cells are filled in that line. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
You are given a grid fully covered by thermometers — straight segments of two or more cells, each with a circular bulb marking one end. Mercury always flows from the bulb and stays contiguous, so each thermometer is filled as a run starting at its bulb. The numbers outside the grid (one per row on the right, one per column on the bottom) state how many cells are filled in that line. Generate a video that fills the cells progressively from the bulbs to show the unique solution. Only modify what the task requires; keep the background and any other element unchanged.
Nurikabe
Image sequence
SFT-Image source
Image prompt
You are given a Nurikabe board with clue numbers (white cells); all remaining cells start gray. Shade the gray cells so that the black sea is fully connected with no 2×2 all-black square, while the white cells form islands — each containing exactly one clue and having as many cells as that clue's value, with no two islands sharing an edge. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
You are given a Nurikabe board with clue numbers (white cells); all remaining cells start gray. Shade the gray cells so that the black sea is fully connected with no 2×2 all-black square, while the white cells form islands — each containing exactly one clue and having as many cells as that clue's value, with no two islands sharing an edge. Generate a video that progressively paints each cell to its final color to show the solution. Only modify what the task requires; keep the background and any other element unchanged.
Shape Outline Then Move
Image sequence
SFT-Image source
Image prompt
The scene shows an analogy A→B→C :: D→?→? with two rows of shapes and arrows. On the top row, a filled T_shape first becomes an outline-only T_shape (step 1), then moves down by a tiny amount (step 2). On the bottom row, the rectangle starts filled. Apply the same two-step transformation: first convert it to outline-only style, then move it down by a tiny amount, keeping its shape and size the same while only the style and position change. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the completed analogy.
Video
Data video source
Video prompt
The scene shows an analogy A→B→C :: D→?→? with two rows of shapes and arrows. On the top row, a filled T_shape first becomes an outline-only T_shape (step 1), then moves down by a tiny amount (step 2). On the bottom row, the rectangle starts filled. Apply the same two-step transformation: first convert it to outline-only style, then move it down by a tiny amount, keeping its shape and size the same while only the style and position change. Only modify what the task requires; keep the background and any other element unchanged.
Numbrix
Image sequence
SFT-Image source
Image prompt
The image shows a 5x5 Numbrix grid. Some cells already contain given numbers (shown in bold), while the remaining cells are empty. The objective is to fill every empty cell so that the grid contains each integer from 1 to 25 exactly once, and so that every pair of consecutive integers occupies cells that are orthogonally adjacent (sharing an edge horizontally or vertically, never diagonally). The completed numbers therefore trace a single continuous path that winds from 1 to 25 through all 25 cells. The given numbers are fixed and the solution is unique. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows a 5x5 Numbrix grid. Some cells already contain given numbers (shown in bold), while the remaining cells are empty. The objective is to fill every empty cell so that the grid contains each integer from 1 to 25 exactly once, and so that every pair of consecutive integers occupies cells that are orthogonally adjacent (sharing an edge horizontally or vertically, never diagonally). The completed numbers therefore trace a single continuous path that winds from 1 to 25 through all 25 cells. The given numbers are fixed and the solution is unique. Generate a video that fills in the empty cells one at a time, in ascending numerical order, until the entire grid is solved. Only modify what the task requires; keep the background and any other element unchanged.
Missing Color Region Fill
Image sequence
SFT-Image source
Image prompt
A 3x3 grid of circles is shown on a white background. Exactly one position is missing and is marked with a '?'. The rule is position-based: the four corners share one color, the four edge-middle cells share one color, and the center uses a third color. Fill the missing circle with the correct solid color, remove the '?', and change nothing else. Output 1 image showing the final grid with the missing circle filled.
Video
Data video source
Video prompt
You are given a 3x3 arrangement of circles. One circle is replaced by a '?'. Circles in the corner positions match each other, circles in the edge-middle positions match each other, and the center has its own color. Infer the correct color for the missing position and fill it in completely while keeping the rest of the scene unchanged.
Shape Rotation
Image sequence
SFT-Image source
Image prompt
The image shows a visual analogy A:B :: C:? using simple shapes. A and C are the original shapes, B shows shape A after a rotation, and ? is the unknown result. Each shape has a small black marker dot to indicate its orientation. Your task is to infer the single rotation that maps A (kite) to B, then apply the same rotation to C (hexagon) around its geometric center. The answer at ? must be shape C rotated by this angle, with size and position unchanged, so that the rotational relationship between A and B is mirrored between C and ?. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The image shows a visual analogy A:B :: C:? using simple shapes. A and C are the original shapes, B shows shape A after a rotation, and ? is the unknown result. Each shape has a small black marker dot to indicate its orientation. Your task is to infer the single rotation that maps A (kite) to B, then apply the same rotation to C (hexagon) around its geometric center. The answer at ? must be shape C rotated by this angle, with size and position unchanged, so that the rotational relationship between A and B is mirrored between C and ?. Only modify what the task requires; keep the background and any other element unchanged.
Draw Next Sized Shape
Image sequence
Data image source
Image prompt
A row of shapes of the same type follows a repeating size cycle. The last box is empty. First determine which size should come next in the cycle, then draw the next shape inside the empty box. Only modify what the task requires; keep the background and any other element unchanged. Output a single image showing the complete final solution.
Video
Data video source
Video prompt
A row of shapes of the same type follows a repeating size cycle. The last box is empty. First determine which size should come next in the cycle, then draw the next shape inside the empty box. Only modify what the task requires; keep the background and any other element unchanged. Show the complete solution step by step.
Hexagon Size Hierarchy Puzzle
Image sequence
SFT-Image source
Image prompt
A 3x3 grid of shapes is shown. Each column uses a fixed shape type, and each row uses a fixed size level (top = large, middle = medium, bottom = small). One shape has its fill color missing and is marked with a '?'. Use the rule: all shapes in the same column must have the same color. Infer the correct color for the '?' shape from the other shapes in its column, then fully fill the '?' shape with that solid color. The '?' is removed. Do not move, resize, rotate, or recolor any other shape. Output 1 image showing the final grid with the '?' shape filled in the correct color.
Video
Data video source
Video prompt
A 3x3 grid of shapes is shown. Each column uses a fixed shape type, and each row uses a fixed size level (top = large, middle = medium, bottom = small). One shape has its fill color missing and is marked with a '?'. Use the rule: all shapes in the same column must have the same color. Infer the correct color for the '?' shape from the other shapes in its column, then fully fill the '?' shape with that solid color. The '?' must disappear. Do not move, resize, rotate, or recolor any other shape. After filling, the scene remains perfectly still.
Select Next Figure Small Large Alternating Sequence
Image sequence
Data image source
Image prompt
The top row shows three figures with the same shape and color; their sizes follow small–large–small left to right. A dashed box marks where the fourth figure belongs. Pick the bottom option that continues the alternating size pattern. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the correct choice marked with a red circle.
Video
Data video source
Video prompt
The top row shows three figures with the same shape and color; their sizes follow small–large–small left to right. A dashed box marks where the fourth figure belongs. Pick the bottom option that continues the alternating size pattern, then the video marks the correct choice with a red circle. Only modify what the task requires; keep the background and any other element unchanged.
Sokoban Minimum Solution Length
Image sequence
SFT-Image source
Image prompt
Solve the Sokoban puzzle optimally under minimum number of moves. Return the minimum cost, and include an optimal path and solved board when available. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the solved board.
Video
Data video source
Video prompt
For the displayed Sokoban board, compute the optimal minimum number of moves; also provide one optimal path and the solved board summary. Only modify what the task requires; keep the background and any other element unchanged.
Symbol Substitution
Image sequence
SFT-Image source
Image prompt
The scene shows a horizontal sequence of colored geometric symbols arranged from left to right. Each symbol is a distinct geometric shape with a specific color, and exactly one symbol is marked as the substitution target by a red rectangular border surrounding it. In symbol substitution tasks, the target symbol identified by the red border must be replaced with a new symbol while all other symbols remain unchanged in their original positions. First identify the symbol marked with the red border, then replace it with a pink triangle. The final state must show the sequence with the target symbol substituted by the pink triangle at the same position, while all other symbols retain their original shapes, colors, and sequential positions. The substitution operation affects only the single marked symbol, replacing it with the pink triangle. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The scene shows a horizontal sequence of colored geometric symbols arranged from left to right. Each symbol is a distinct geometric shape with a specific color, and exactly one symbol is marked as the substitution target by a red rectangular border surrounding it. In symbol substitution tasks, the target symbol identified by the red border must be replaced with a new symbol while all other symbols remain unchanged in their original positions. First identify the symbol marked with the red border, then replace it with a pink triangle. The final state must show the sequence with the target symbol substituted by the pink triangle at the same position, while all other symbols retain their original shapes, colors, and sequential positions. The substitution operation affects only the single marked symbol, replacing it with the pink triangle. Only modify what the task requires; keep the background and any other element unchanged.
Find Fragment For Gap Filling
Image sequence
SFT-Image source
Image prompt
The scene has two separated areas: a top PUZZLE area and a bottom CHOICES area. In the PUZZLE area, the center shape has a missing cut-out outlined in black. In the CHOICES area, there are 4 candidate pieces of the same shape but different sizes. First compare the candidate sizes and determine which single option would fit the missing cut-out exactly (a perfect match in size). Do not use color as a clue. Then circle the correct option with a red circle. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The scene has two separated areas: a top PUZZLE area and a bottom CHOICES area. In the PUZZLE area, the center shape has a missing cut-out outlined in black. In the CHOICES area, there are 4 candidate pieces of the same shape but different sizes. First compare the candidate sizes and determine which single option would fit the missing cut-out exactly (a perfect match in size). Do not use color as a clue. Then circle the correct option with a red circle. Show the complete solution step by step. Only modify what the task requires; keep the background and any other element unchanged.
Match3 Jewel Swap Result
Image sequence
SFT-Image source
Image prompt
You are given a square match-3 jewel board (rows and columns labeled from the top-left). There is exactly one pair of orthogonally adjacent jewels whose swap creates at least one straight line of three or more identical jewels—no other adjacent swap does. Identify that swap as (row1,col1)↔(row2,col2) using the shown coordinates. Then state which jewel type(s) clear and the immediate elimination: all cleared cells and the total count after that swap, before gravity or refills. Only modify what the task requires; keep the background and any other element unchanged. Output 2 images showing the swap and the resulting elimination at uniform stages.
Video
Data video source
Video prompt
You are given a square match-3 jewel board (rows and columns labeled from the top-left). There is exactly one pair of orthogonally adjacent jewels whose swap creates at least one straight line of three or more identical jewels—no other adjacent swap does. Identify that swap as (row1,col1)↔(row2,col2) using the shown coordinates. Then state which jewel type(s) clear and the immediate elimination: all cleared cells and the total count after that swap, before gravity or refills. Only modify what the task requires; keep the background and any other element unchanged.
Arrow And Hexagon Direction Puzzle
Image sequence
SFT-Image source
Image prompt
What is the color of the missing part denoted with a question mark? Infer the missing cell's fill color from the repeating pattern in the triangular pyramid of arrows and hexagons (its bands follow a consistent rule along diagonals or rows). This part should be completely filled with the correct solid color while the other parts should be unchanged. The question mark is removed. Output 1 image showing the final pyramid with the missing cell filled.
Video
Data video source
Video prompt
The pyramid follows a consistent rule along its bands (diagonals or rows, depending on the sample). Determine which solid color belongs in the marked cell, fill it completely, and leave every other pixel of the layout unchanged. Remove the question mark and keep the scene static.
Tapa
Image sequence
SFT-Image source
Image prompt
Tapa puzzle: shade cells so they form one connected region with no 2×2 area fully shaded. Every clue cell is left unshaded, and its numbers give the sizes of the separate blocks of consecutive shaded cells among its eight surrounding cells; multiple numbers mean multiple separate blocks (in any order). Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Tapa puzzle: shade cells so they form one connected region with no 2×2 area fully shaded. Every clue cell is left unshaded, and its numbers give the sizes of the separate blocks of consecutive shaded cells among its eight surrounding cells; multiple numbers mean multiple separate blocks (in any order). Generate a video that reveals the correct shading until the puzzle is solved. Only modify what the task requires; keep the background and any other element unchanged.
Sokoban
Image sequence
SFT-Image source
Image prompt
Solve the Sokoban puzzle in the image by pushing each box onto a target. Boxes can only be pushed, one at a time, and walls block movement. For each box in turn, output two images: (1) the moment the player reaches that box, just before pushing it, and (2) that box pushed onto its goal. In each image an orange line marks the player's route into that state — the walk up to the box in the first image, and the pushing path that drives the box onto its goal in the second. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Solve the Sokoban puzzle in the image by pushing each box onto a target. Boxes can only be pushed, one at a time, and walls block movement. Generate a video that plays the solution move by move, ending with all boxes on their goals. Only modify what the task requires; keep the background and any other element unchanged.
Predict Next Color
Image sequence
SFT-Image source
Image prompt
Predict the next color in the sequence. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
Predict the next color in the sequence. Only modify what the task requires; keep the background and any other element unchanged.
Hanoi
Image sequence
SFT-Image source
Image prompt
The image shows a Tower of Hanoi puzzle with 3 disks on 3 pegs (labelled 1–3 from left to right). Move all disks onto peg 3 so they end largest-on-bottom to smallest-on-top, following the rules: only the top disk of a peg may move, one disk at a time, and a larger disk may never rest on a smaller one. Output 5 images showing the tower after each move, in order. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows a Tower of Hanoi puzzle with 3 disks on 3 pegs (labelled 1–3 from left to right). Move all disks onto peg 3 so they end largest-on-bottom to smallest-on-top, following the rules: only the top disk of a peg may move, one disk at a time, and a larger disk may never rest on a smaller one. Generate a video that performs the optimal sequence of moves one at a time until every disk is stacked on peg 3. Only modify what the task requires; keep the background and any other element unchanged.
Symbol Insert
Image sequence
Data image source
Image prompt
Insert a orange hollow diamond at position 5. A reference panel in the top-right shows the target symbol. Existing symbols to the right of the target slot slide right to make room, then the new symbol fades in above the gap and slides down into the target slot. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
Insert a orange hollow diamond at position 5. A reference panel in the top-right shows the target symbol. Existing symbols to the right of the target slot slide right to make room, then the new symbol fades in above the gap and slides down into the target slot. Only modify what the task requires; keep the background and any other element unchanged.
Overlay Patterns Result
Image sequence
SFT-Image source
Image prompt
Two distinct sparse patterns are shown separately on the same cell layout. Determine the final visual result after overlaying both patterns, where any occupied cell in either layer appears in the combined result. Use the top examples to identify the missing merged panel and choose the correct option among A/B/C. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final overlaid result with the correct option marked.
Video
Data video source
Video prompt
Two distinct sparse patterns are shown separately on the same cell layout. Determine the final visual result after overlaying both patterns, where any occupied cell in either layer appears in the combined result. Only modify what the task requires; keep the background and any other element unchanged.
Overlap Shape Matching
Image sequence
SFT-Image source
Image prompt
What is the color of the missing part denoted with a question mark? The mark may appear on the left, middle, or right band. That region must be completely filled with the correct color while every other region stays unchanged. The question mark is removed. Overlapping colored shapes use simple color mixing in their intersections (average of the two layer colors); use the overlap regions as hints. Output 1 image showing the final scene with the missing band filled.
Video
Data video source
Video prompt
What is the color of the missing part denoted with a question mark? The mark may appear on the left, middle, or right band. That region must be completely filled with the correct color while every other region stays unchanged. The question mark disappears, then the scene remains static with no zoom and no movement of the shapes.
Rotation Pattern Prediction
Image sequence
SFT-Image source
Image prompt
A 1024×1024 panel shows a regular polygonal ring of colored sectors. Three snapshots show the same pattern advanced by a fixed discrete rotation; a gray circle marks the missing next frame. Options A–C propose successor rings—only one continues the same rule. A pivot, curved arrow, and angle text encode the step. Infer the discrete rotation and select the next orientation. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final ring with the correct option highlighted.
Video
Data video source
Video prompt
A panel shows a regular polygonal ring of colored sectors (segment count, palette, and hues vary by task). Three snapshots show the same pattern advanced by a fixed discrete rotation; a gray circle marks the missing next frame. Options A–C propose successor rings—only one continues the same rule. A pivot, curved arrow, and angle text encode the step. The supervised video ends with the correct ring and highlighted option. Only modify what the task requires; keep the background and any other element unchanged.
Space Invaders Shooting Timing
Image sequence
SFT-Image source
Image prompt
A Space Invaders scene on a fixed 7x7 grid: a player ship on the bottom row with aliens above. Each step the player can move LEFT, move RIGHT, SHOOT (a bullet goes straight up), or STAY.
The target alien is the lowest one (largest row index); if several share that row, the leftmost. Find the SHORTEST action sequence that destroys the target, accounting for movement and the upward bullet.
Draw the completed plan on a single image: an arrow along the bottom row showing the ship's movement to the firing column, then an upward arrow showing the bullet reaching the target alien (marked destroyed), with the whole action sequence captioned below the grid, for example: Actions: RIGHT → RIGHT → SHOOT. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the completed plan.
Video
Data video source
Video prompt
A Space Invaders scene on a fixed 7x7 grid: a player ship sits on the bottom row with aliens above. Each step the player can move LEFT (one cell left), move RIGHT (one cell right), SHOOT (fire a bullet straight up), or STAY.
The target alien is the lowest one (largest row index); if several share that row, the leftmost. Find the SHORTEST action sequence that destroys the target, accounting for the ship's movement and the upward bullet trajectory.
Generate a video of the player carrying out that sequence: it steps into the target's column, then fires; the bullet rises straight up and the target alien explodes. On the final frames, write the action sequence as a caption below the grid, for example: Actions: RIGHT → RIGHT → SHOOT. End on the board with the target removed. Only modify what the task requires; keep the background and any other element unchanged.
Nested Pentagon Center Target
Image sequence
SFT-Image source
Image prompt
The missing center part is marked with a '?'. A nested regular polygon (e.g. triangle, square, pentagon, hexagon) has a missing center color shown as '?'. Predict the correct fill from the visible concentric color gradient, fill the innermost region with that color, and keep all outer concentric rings unchanged. The question mark is removed. Output 1 image showing the final figure with the center filled.
Video
Data video source
Video prompt
The missing center part is marked with a '?'. Fill that innermost region with the correct color while keeping all outer concentric rings unchanged.
Space Invaders Evasion Path Prediction
Image sequence
SFT-Image source
Image prompt
A Space Invaders scene on a fixed 7x7 grid: the player ship on the bottom row (row 7); enemy bullets above fall straight down one cell per step. Each step the player may move LEFT, move RIGHT, or STAY. A collision happens if a bullet ever shares the player's cell.
Find the safe action sequence of exactly 5 steps that dodges every bullet, using the FEWEST moves; if several tie, the lexicographically smallest sequence. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image: the player's final position after the 5 steps (with the bullets descended), and the full action sequence captioned below the grid.
Video
Data video source
Video prompt
A Space Invaders scene on a fixed 7x7 grid: the player ship sits on the bottom row (row 7); enemy bullets above fall straight down one cell per step. Each step the player may move LEFT (one cell left), move RIGHT (one cell right), or STAY. A collision happens if a bullet ever shares the player's cell.
Find the safe action sequence of exactly 5 steps that dodges every bullet, using the FEWEST moves; if several tie, the lexicographically smallest sequence.
Generate a video of the player executing that sequence over the 5 steps as the bullets descend, never being hit. On the final frames, write the chosen sequence as a caption below the grid, for example: Actions: LEFT → LEFT → STAY → RIGHT → STAY. Only modify what the task requires; keep the background and any other element unchanged.
Langtons Ant Final State Prediction
Image sequence
SFT-Image source
Image prompt
Langton's Ant rule: on a white cell, turn 90° right and flip that cell to black; on a black cell, turn 90° left and flip that cell to white; then move forward one cell. Simulate Langton's Ant from the shown initial board and draw the final board state. The final state shows the ant's final cell, its facing direction, and each highlighted key cell's color. Run it for the number of steps shown on the board. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board state.
Video
Data video source
Video prompt
Langton's Ant rule: on a white cell, turn 90° right and flip that cell to black; on a black cell, turn 90° left and flip that cell to white; then move forward one cell. Run Langton's Ant from the shown starting board step by step until it reaches the final state. The final state shows the ant's final cell, its facing direction, and each highlighted key cell's color. Run it for the number of steps shown on the board. Only modify what the task requires; keep the background and any other element unchanged.
Shape Matrix Missing Piece
Image sequence
SFT-Image source
Image prompt
Two rows of shapes are shown, separated by a horizontal mirror line. The top row is the vertical reflection of the bottom row across that line (same x positions, mirrored in y). One top shape is missing its fill color and is marked with a '?'. Use the rule: a reflected pair (top and bottom at the same column) must have the same color. Infer the correct color for the '?' shape from its reflected counterpart, then fully fill the '?' shape with that solid color. The '?' is removed. Do not move, resize, rotate, or recolor any other shape. Output 1 image showing the final scene with the '?' shape filled in the correct color.
Video
Data video source
Video prompt
Two rows of shapes are shown, separated by a horizontal mirror line. The top row is the vertical reflection of the bottom row across that line (same x positions, mirrored in y). One top shape is missing its fill color and is marked with a '?'. Use the rule: a reflected pair (top and bottom at the same column) must have the same color. Infer the correct color for the '?' shape from its reflected counterpart, then fully fill the '?' shape with that solid color. The '?' must disappear. Do not move, resize, rotate, or recolor any other shape. After filling, the scene remains perfectly still.
Select Next Figure Increasing Size Sequence
Image sequence
SFT-Image source
Image prompt
The scene has two separated areas: a top SEQUENCE area and a bottom CHOICES area. In the SEQUENCE area, the shapes are the same shape and the same color, and their sizes strictly increase from left to right. First identify the constant size step between consecutive sequence shapes, then select the one correct option (out of 4) in the CHOICES area that continues the same shape, color, and size-increase pattern. Circle the correct option with a red circle. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
The scene has two separated areas: a top SEQUENCE area and a bottom CHOICES area. In the SEQUENCE area, the shapes are the same shape and the same color, and their sizes strictly increase from left to right. First identify the constant size step between consecutive sequence shapes, then select the one correct option (out of 4) in the CHOICES area that continues the same shape, color, and size-increase pattern. Circle the correct option with a red circle and show the full process step by step. Only modify what the task requires; keep the background and any other element unchanged.
Sliding Puzzle
Image sequence
SFT-Image source
Image prompt
Complete this sliding puzzle. The goal is to arrange the numbered tiles in sequential order (filling each row from left to right, with rows from top to bottom), with the blank space at the bottom-right corner. Rules: Only tiles adjacent to the blank space can be moved. Slide one tile per move into the blank space. Complete in exactly 7 moves. Do not make extra moves. Keep the camera view fixed and maintain the grid structure unchanged. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each move, showing the board state after the tile has been slid into the blank space.
Video
Data video source
Video prompt
Complete this sliding puzzle. The goal is to arrange the numbered tiles in sequential order (filling each row from left to right, with rows from top to bottom), with the blank space at the bottom-right corner.
Rules: Only tiles adjacent to the blank space can be moved. Slide one tile per move into the blank space.
Complete in exactly 7 moves.
Do not make extra moves. Keep the camera view fixed and maintain the grid structure unchanged. Only modify what the task requires; keep the background and any other element unchanged.
Color Hue Relative Ordering
Image sequence
SFT-Image source
Image prompt
A 5x5 grid with black borders and labelled rows and columns holds several solid tiles of similar shade arranged in a single connected chain whose hue increases smoothly from one end to the other, joined by a line drawn through them in hue order with a round dot on the start tile and an arrow on the end tile. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the tiles settled into the connected hue-ordered chain with the line drawn through them.
Video
Data video source
Video prompt
Animate the sorting of coloured tiles on a 5x5 lattice. Beginning from a scrambled layout, the tiles step one by one through the empty checkerboard cells and settle into a connected chain along which the hue changes steadily in one direction. A connecting line is then drawn across the chain in hue order, starting from a dot on the first tile and ending in an arrow on the last. The closing frame holds the tidy, hue-ordered chain with its line. Only modify what the task requires; keep the background and any other element unchanged.
Tents Tree Constraint Completion
Image sequence
SFT-Image source
Image prompt
Tents & Trees on a grid: the green icons are trees. Place tents so that (1) every tree has exactly one tent in an orthogonally adjacent cell and each tent pairs with one tree; (2) no two tents touch, even diagonally; and (3) the number beside each row and above each column equals how many tents that row or column holds. The puzzle has a unique solution. For each highlighted query cell, draw a tent if it must contain one or an X if it can never contain one; leave the rest of the board as the puzzle. Output 1 image with each highlighted cell resolved to a tent or an X.
Video
Data video source
Video prompt
Tents & Trees on a grid: the green icons are trees. Place tents so that (1) every tree has exactly one tent in an orthogonally adjacent cell and each tent pairs with one tree; (2) no two tents touch, even diagonally; and (3) the number beside each row and above each column equals how many tents that row or column holds. The puzzle has a unique solution. You do not need to place every tent. For each highlighted query cell, draw a tent on it if a tent is forced there, or an X on it if a tent is impossible there; leave the rest of the board as the puzzle.
24 Points
Image sequence
SFT-Image source
Image prompt
The image shows four playing-card style tiles, each displaying a number. Use each of the four shown numbers exactly once, combining them with the operations addition (+), subtraction (−), multiplication (×) and division (÷) together with parentheses, to build a single arithmetic expression whose value is exactly 24. The goal is to find one such valid expression for the four numbers on the cards. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows four playing-card style tiles, each displaying a number. Use each of the four shown numbers exactly once, combining them with the operations addition (+), subtraction (−), multiplication (×) and division (÷) together with parentheses, to build a single arithmetic expression whose value is exactly 24. The goal is to find one such valid expression for the four numbers on the cards. Generate a video that presents the four cards and then reveals a correct expression equal to 24. Only modify what the task requires; keep the background and any other element unchanged.
Shape Outline Fill
Image sequence
Data image source
Image prompt
Complete the A:B :: C:? shape-style analogy. Show how the right shape in the second row changes its fill or outline so that it follows the same style transformation used between the first two shapes. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the completed analogy.
Video
Data video source
Video prompt
Complete the A:B :: C:? shape-style analogy. Show how the right shape in the second row changes its fill or outline so that it follows the same style transformation used between the first two shapes. Only modify what the task requires; keep the background and any other element unchanged.
Shape Color Then Move
Image sequence
Data image source
Image prompt
The scene shows a sequential analogy A→B→C :: D→?→? with three shapes per row. On the top row, a diamond changes its color, then moves up. On the bottom row, apply the same two-step transformation to the square: first recolor it to match the new color demonstrated on the top row, then move up. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the completed analogy.
Video
Data video source
Video prompt
The scene shows a sequential analogy A→B→C :: D→?→? with three shapes per row. On the top row, a diamond changes its color, then moves up. On the bottom row, apply the same two-step transformation to the square: first recolor it to match the new color demonstrated on the top row, then move up. Show both transformations sequentially in the video. Only modify what the task requires; keep the background and any other element unchanged.
Shape Color Then Scale
Image sequence
SFT-Image source
Image prompt
The scene shows an analogy A→B→C :: D→?→? with two rows of shapes. Top row: color_324 diamond at medium becomes color_319 diamond at medium, then color_319 diamond at large. This shows two rules: first change color from color_324 to color_319, then change size from medium to large. Bottom row: square starts with color_324 and medium. Apply the same transformation: first change color to color_319, then change size to large. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the completed analogy.
Video
Data video source
Video prompt
The scene shows an analogy A→B→C :: D→?→? with two rows of shapes. Top row: color_324 diamond at medium becomes color_319 diamond at medium, then color_319 diamond at large. This shows two rules: first change color from color_324 to color_319, then change size from medium to large. Bottom row: square starts with color_324 and medium. Apply the same transformation: first change color to color_319, then change size to large. Only modify what the task requires; keep the background and any other element unchanged.
Metro Map Path Trace
Image sequence
SFT-Image source
Image prompt
Static square 1024×1024 schematic metro map: several color-coded lines meet at shared transfer hubs. A start station and a destination station are marked. Find the shortest (fewest-stops) walk along edges and draw a line tracing the full route, highlighting each segment and ringing every interchange where the rider switches color lines. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final map with the completed route drawn.
Video
Data video source
Video prompt
Static square 1024×1024 schematic metro map: several color-coded lines meet at shared transfer hubs. A start station and a destination station are marked. Animate the full shortest (fewest-stops) walk along edges, cumulatively highlighting each segment and ringing every interchange where the rider switches color lines. End with the entire route visible and readable. Only modify what the task requires; keep the background and any other element unchanged.
Battleships
Image sequence
SFT-Image source
Image prompt
Classic Solitaire Battleships with fleet 1x3, 2x2, 2x1 (read as count x length): using the row/column tallies and the revealed hint cells (a dark square must be covered by a ship, an × must stay empty water), drop each ship so the fleet fits without any two ships adjacent. Animate the placement of the ships and hold on the final frame showing the entire fleet correctly placed. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Classic Solitaire Battleships with fleet 1x3, 2x2, 2x1 (read as count x length): using the row/column tallies and the revealed hint cells (a dark square must be covered by a ship, an × must stay empty water), drop each ship so the fleet fits without any two ships adjacent. Animate the placement of the ships and hold on the final frame showing the entire fleet correctly placed. Only modify what the task requires; keep the background and any other element unchanged.
Aquarium
Image sequence
SFT-Image source
Image prompt
The image shows an Aquarium logic puzzle on a square grid that is divided into irregular regions ("aquariums") outlined by thick black borders; thin lines mark the individual cells. The number to the right of each row and the number below each column is a clue giving exactly how many cells in that row or column must contain water. Water obeys gravity: within every region it settles to a single flat level and fills the region from the bottom of the grid upward, so a cell is filled only if every cell of the same region directly below it (in the same column) is also filled — there is no floating or stepped water inside a region. The goal is to choose a water level for each region so that all row clues and all column clues are satisfied simultaneously. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows an Aquarium logic puzzle on a square grid that is divided into irregular regions ("aquariums") outlined by thick black borders; thin lines mark the individual cells. The number to the right of each row and the number below each column is a clue giving exactly how many cells in that row or column must contain water. Water obeys gravity: within every region it settles to a single flat level and fills the region from the bottom of the grid upward, so a cell is filled only if every cell of the same region directly below it (in the same column) is also filled — there is no floating or stepped water inside a region. The goal is to choose a water level for each region so that all row clues and all column clues are satisfied simultaneously. Generate a video that fills the correct cells with water, starting from the bottom row and working upward, until the completed solution is shown. Only modify what the task requires; keep the background and any other element unchanged.
Dot Grid Size Pattern Completion
Image sequence
SFT-Image source
Image prompt
A 3x3 grid of circles is shown. Each circle belongs to a size category based on its radius. Use the rule: circles with the same size category must have the same color. One circle is missing its fill color and is marked with a '?'. Infer the correct color by matching its size category to other circles of the same size, then fully fill the missing circle with that solid color. The '?' must disappear. Do not move, resize, or recolor any other circle. Output 1 image showing the final grid with the missing circle filled.
Video
Data video source
Video prompt
A 3x3 grid of circles is shown. Each circle belongs to a size category based on its radius. Use the rule: circles with the same size category must have the same color. One circle is missing its fill color and is marked with a '?'. Infer the correct color by matching its size category to other circles of the same size, then fully fill the missing circle with that solid color. The '?' must disappear. Do not move, resize, or recolor any other circle. After filling, the scene remains perfectly still.
Sokoban State Prediction
Image sequence
SFT-Image source
Image prompt
A Sokoban board is shown with the move sequence DLRLRLDUD listed at the top. Generate a video where the player carries out the moves one at a time, walking and pushing boxes; a move blocked by a wall or an immovable box leaves everything in place (the cell it could not enter is briefly outlined). The board updates step by step, ending with the player at its final cell. The camera stays fixed. Only modify what the task requires; keep the background and any other element unchanged. Output 9 images showing the board after each move at uniform stages.
Video
Data video source
Video prompt
A Sokoban board is shown with the move sequence DLRLRLDUD listed at the top. Generate a video where the player carries out the moves one at a time, walking and pushing boxes; a move blocked by a wall or an immovable box leaves everything in place (the cell it could not enter is briefly outlined). The board updates step by step, ending with the player at its final cell. The camera stays fixed. Only modify what the task requires; keep the background and any other element unchanged.
Raven
Image sequence
Data image source
Image prompt
Complete the progressive matrix puzzle: infer the rule from the eight visible cells and draw the correct pattern in the bottom-right cell. Do not alter any other cell. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
Complete the progressive matrix puzzle: infer the rule from the eight visible cells and draw the correct pattern in the bottom-right cell. Do not alter any other cell. Only modify what the task requires; keep the background and any other element unchanged.
Select Next Figure Large Small Alternating Sequence
Image sequence
SFT-Image source
Image prompt
The scene has two separated areas: a top SEQUENCE area and a bottom CHOICES area. In the SEQUENCE area, the shapes are the same shape and the same color, and their sizes strictly alternate between LARGE and SMALL from left to right. First observe the size-alternation pattern and determine whether the next item should be LARGE or SMALL, then select the one correct option (out of 4) in the CHOICES area that continues the same shape, color, and large/small alternation pattern. Circle the correct option with a red circle. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The scene has two separated areas: a top SEQUENCE area and a bottom CHOICES area. In the SEQUENCE area, the shapes are the same shape and the same color, and their sizes strictly alternate between LARGE and SMALL from left to right. First observe the size-alternation pattern and determine whether the next item should be LARGE or SMALL, then select the one correct option (out of 4) in the CHOICES area that continues the same shape, color, and large/small alternation pattern. Circle the correct option with a red circle and show the full process step by step. Only modify what the task requires; keep the background and any other element unchanged.
Sliding Number Puzzle Sort
Image sequence
SFT-Image source
Image prompt
A 3x3 sliding number puzzle with tiles 1-8 and one empty cell is shown. Slide the numbered tiles into the empty space so that the numbers are in ascending order from 1 to 8, with the empty cell in the bottom-right corner. Output one image after each sliding move, showing the full sequence of valid moves from the scrambled start to the solved configuration. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
A 3x3 sliding number puzzle with tiles 1-8 and one empty cell is shown. Slide the numbered tiles into the empty space so that the numbers are in ascending order from 1 to 8, with the empty cell in the bottom-right corner. The animation must demonstrate the full sequence of valid sliding moves from the scrambled start configuration to the solved configuration.
Sudoku Color Logic Deduction
Image sequence
SFT-Image source
Image prompt
A 4x4 color Sudoku with a four-color legend at the top; each row, column, and 2x2 box must contain all four colors exactly once. Complete the grid by filling every empty cell with its correct color. Output 1 image showing the fully completed grid.
Video
Data video source
Video prompt
A 4x4 color Sudoku. The legend at the top shows the four colors; each row, column, and 2x2 box must contain all four exactly once. Fill every empty cell with its correct color to complete the grid.
Shikaku
Image sequence
SFT-Image source
Image prompt
The image shows a grid with a number in some of its cells. Divide the entire grid into axis-aligned rectangles so that the rectangles do not overlap and leave no cell uncovered. Each rectangle must contain exactly one number, and that number must equal the rectangle's area (its cell count). The numbers yield a unique partition; generate a video that reveals the rectangles one at a time until the whole grid is divided. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows a grid with a number in some of its cells. Divide the entire grid into axis-aligned rectangles so that the rectangles do not overlap and leave no cell uncovered. Each rectangle must contain exactly one number, and that number must equal the rectangle's area (its cell count). The numbers yield a unique partition; generate a video that reveals the rectangles one at a time until the whole grid is divided. Only modify what the task requires; keep the background and any other element unchanged.
Kakuro
Image sequence
SFT-Image source
Image prompt
Solve this Kakuro puzzle from the image: enter digits 1–9 into the white cells so each run adds up to its clue with no repeated digit in a run. In a black clue cell, the number above the diagonal is the sum for the horizontal run of white cells to its right, and the number below the diagonal is the sum for the vertical run of white cells below it. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Solve this Kakuro puzzle from the image: enter digits 1–9 into the white cells so each run adds up to its clue with no repeated digit in a run. In a black clue cell, the number above the diagonal is the sum for the horizontal run of white cells to its right, and the number below the diagonal is the sum for the vertical run of white cells below it. Generate a video that reveals the correct digits until the whole grid is filled. Only modify what the task requires; keep the background and any other element unchanged.
Ball Eating
Image sequence
SFT-Image source
Image prompt
In the scene, there is a black ball and several colored balls of different sizes. The black ball can eat balls that are smaller than itself. After eating a ball, the black ball grows larger. Find the correct sequence to eat all colored balls step by step. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each eating event, showing the updated state.
Video
Data video source
Video prompt
In the scene, there is a black ball and several colored balls of different sizes. The black ball can eat balls that are smaller than itself. After eating a ball, the black ball grows larger. Find the correct sequence to eat all colored balls step by step. Only modify what the task requires; keep the background and any other element unchanged.
Shingoki
Image sequence
SFT-Image source
Image prompt
Solve this Shingoki puzzle: form one continuous non-crossing loop passing through all the circles, going straight at white circles and turning at black ones, with each number matching the summed length of straight lines from that circle. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Solve this Shingoki puzzle: form one continuous non-crossing loop passing through all the circles, going straight at white circles and turning at black ones, with each number matching the summed length of straight lines from that circle. Generate a video that traces the loop step by step until it is complete. Only modify what the task requires; keep the background and any other element unchanged.
Symbol Edit
Image sequence
Data image source
Image prompt
The sequence currently has 1 of symbol hollow circle. Constraint: at least 4 of symbol hollow circle. A reference panel in the top-right shows the target symbol. Insert 3 hollow circle symbols at positions 1, 5, and 6 to satisfy the constraint. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The sequence currently has 1 of symbol hollow circle. Constraint: at least 4 of symbol hollow circle. A reference panel in the top-right shows the target symbol. Insert 3 hollow circle symbols at positions 1, 5, and 6 to satisfy the constraint. Only modify what the task requires; keep the background and any other element unchanged.
Lightup
Image sequence
SFT-Image source
Image prompt
You are given a Light Up puzzle. Black walls block light, and a wall labeled with a number requires exactly that many bulbs in its orthogonally adjacent cells (0 means none). Put bulbs on white cells so that every white cell is lit by some bulb in its row or column (light stops at walls), no two bulbs can see each other along an open line, and all wall numbers are met. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
You are given a Light Up puzzle. Black walls block light, and a wall labeled with a number requires exactly that many bulbs in its orthogonally adjacent cells (0 means none). Put bulbs on white cells so that every white cell is lit by some bulb in its row or column (light stops at walls), no two bulbs can see each other along an open line, and all wall numbers are met. Generate a video that places one bulb at a time and shades the newly lit cells until the grid is solved. Only modify what the task requires; keep the background and any other element unchanged.
Shape Color Change
Image sequence
SFT-Image source
Image prompt
The scene shows an analogy A:B :: C:? with two rows of shapes and arrows. On the top row, a purple diamond becomes a olive diamond; treat this as the color rule from purple to olive. On the bottom row, the square starts with the same source color. Apply the same color change so the answer shape on the right is a olive square, keeping its shape, size, and position the same while only the color changes. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the completed analogy.
Video
Data video source
Video prompt
The scene shows an analogy A:B :: C:? with two rows of shapes and arrows. On the top row, a purple diamond becomes a olive diamond; treat this as the color rule from purple to olive. On the bottom row, the square starts with the same source color. Apply the same color change so the answer shape on the right is a olive square, keeping its shape, size, and position the same while only the color changes. Only modify what the task requires; keep the background and any other element unchanged.
Star Battle Valid Star Placement
Image sequence
SFT-Image source
Image prompt
Star Battle (one star per row, column, and colored region; stars cannot touch, even diagonally). Given the partially filled grid and region shading, decide where the next star must legally go. Name that cell (e.g. row–column in reading order from the top-left) and show it placed in the final state. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the grid with the next star placed.
Video
Data video source
Video prompt
Puzzle: colored regions define groups; exactly one star lives in each row, each column, and each group; two stars cannot touch at a corner or edge. Given current black dots, which single empty cell must receive the next star? Identify it and animate or illustrate the completed board. Only modify what the task requires; keep the background and any other element unchanged.
Select Next Figure Decreasing Size Sequence
Image sequence
SFT-Image source
Image prompt
The scene has two separated areas: a top SEQUENCE area and a bottom CHOICES area. In the SEQUENCE area, the shapes are the same shape and the same color, and their sizes strictly decrease from left to right. First identify the constant size step between consecutive sequence shapes, then select the one correct option (out of 4) in the CHOICES area that continues the same shape, color, and size-decrease pattern. Circle the correct option with a red circle. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The scene has two separated areas: a top SEQUENCE area and a bottom CHOICES area. In the SEQUENCE area, the shapes are the same shape and the same color, and their sizes strictly decrease from left to right. First identify the constant size step between consecutive sequence shapes, then select the one correct option (out of 4) in the CHOICES area that continues the same shape, color, and size-decrease pattern. Circle the correct option with a red circle and show the full process step by step. Only modify what the task requires; keep the background and any other element unchanged.
Pac-Man Safe Collection Route
Image sequence
SFT-Image source
Image prompt
Given the image of a Pac-Man safe-collection puzzle, infer the full grid state from the visible maze: walls, walkable cells, pellets, Pac-Man, and ghosts. Use a planning horizon of exactly 4 moves. Ghost mode is 'static_risk'. Under 'static_risk', ghosts stay at their shown cells and only create danger fields; under 'scripted' or 'chase', ghost movement must follow the fixed generator rule implied by that mode. Treat strict no-adjacent ghost safety as no. Allow waiting in place: no. Enumerate all legal action sequences of that exact horizon, score them deterministically, and rank them by this fixed order: survive the horizon first, then collect more pellets, then higher route score, then shorter route length, then lexicographic route order. The route score itself is deterministic: reward collected pellets, penalize danger, adjacency to ghosts, waiting, and backtracking, then add escape-space preference and target-pellet proximity exactly according to the generator rules. Output only the immediate action, the best sequence, and the target pellet implied by this deterministic short-horizon policy. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final board after the best route is followed, with collected pellets removed.
Video
Data video source
Video prompt
Given the first frame of a Pac-Man safe-collection puzzle, infer the full grid state from the visible maze: walls, walkable cells, pellets, Pac-Man, and ghosts. Use a planning horizon of exactly 4 moves. Ghost mode is 'static_risk'. Under 'static_risk', ghosts stay at their shown cells and only create danger fields; under 'scripted' or 'chase', ghost movement must follow the fixed generator rule implied by that mode. Treat strict no-adjacent ghost safety as no. Allow waiting in place: no. Enumerate all legal action sequences of that exact horizon, score them deterministically, and rank them by this fixed order: survive the horizon first, then collect more pellets, then higher route score, then shorter route length, then lexicographic route order. The route score itself is deterministic: reward collected pellets, penalize danger, adjacency to ghosts, waiting, and backtracking, then add escape-space preference and target-pellet proximity exactly according to the generator rules. The video must be uniquely determined from this first frame and these rules: show the initial board, then the danger overlay, then the top candidate routes under that ranking, then replay only the selected best route with pellets disappearing only when Pac-Man actually reaches them, and finally hold on that finished replay state. Output only the immediate action, the best sequence, and the target pellet implied by this deterministic short-horizon policy. Only modify what the task requires; keep the background and any other element unchanged.
Turing Machine Grid Transition
Image sequence
SFT-Image source
Image prompt
A 2D Turing machine runs on the coloured grid: each cell's colour is its symbol, and the black dot is the read/write head, sitting on one cell at a (row, column) position. The machine is also in one of several states named q1, q2, q3, … — a state is the machine's internal mode, NOT a position on the grid. Each transition rule is written 'state, read-colour -> write-colour, move, next-state' and means: when the machine is in that state and the head reads that colour, it writes the new colour into the head's current cell, moves the head one cell (U/D/L/R = up/down/left/right), and switches to the next state. The board is finite: a move that would leave the grid is not taken and the machine stops. Start from the state and head position shown and run the number of steps given in the header. Report the final state (qN), the final head position (row, column), and a short summary of the writes and moves. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the machine after each step until the end.
Video
Data video source
Video prompt
A 2D Turing machine runs on the coloured grid: each cell's colour is its symbol, and the black dot is the read/write head, sitting on one cell at a (row, column) position. The machine is also in one of several states named q1, q2, q3, … — a state is the machine's internal mode, NOT a position on the grid. Each transition rule is written 'state, read-colour -> write-colour, move, next-state' and means: when the machine is in that state and the head reads that colour, it writes the new colour into the head's current cell, moves the head one cell (U/D/L/R = up/down/left/right), and switches to the next state. The board is finite: a move that would leave the grid is not taken and the machine stops. Start from the state and head position shown and run the number of steps given in the header. Report the final state (qN), the final head position (row, column), and a short summary of the writes and moves. Only modify what the task requires; keep the background and any other element unchanged.
Symbol Deletion
Image sequence
Data image source
Image prompt
The scene shows a horizontal sequence of colored geometric symbols arranged from left to right. Each symbol is a distinct geometric shape with a specific color, and exactly one symbol is marked as the deletion target by a red rectangular border surrounding it. In symbol deletion tasks, the target symbol identified by the red border must be deleted from the sequence while all remaining symbols maintain their original sequential order. First identify the symbol marked with the red border, then delete it from the sequence. The final state must show the remaining symbols in their original order, with the target symbol completely removed. The deletion operation affects only the single marked symbol, leaving all other symbols unchanged in their shapes, colors, and sequential positions. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The scene shows a horizontal sequence of colored geometric symbols arranged from left to right. Each symbol is a distinct geometric shape with a specific color, and exactly one symbol is marked as the deletion target by a red rectangular border surrounding it. In symbol deletion tasks, the target symbol identified by the red border must be deleted from the sequence while all remaining symbols maintain their original sequential order. First identify the symbol marked with the red border, then delete it from the sequence. The final state must show the remaining symbols in their original order, with the target symbol completely removed. The deletion operation affects only the single marked symbol, leaving all other symbols unchanged in their shapes, colors, and sequential positions. Only modify what the task requires; keep the background and any other element unchanged.
Logic Patterns Completion
Image sequence
SFT-Image source
Image prompt
A 5×5 grid shows symbolic cells following a fixed rule: the two outer columns are mirror-identical; each outer cell is split into three horizontal bands with one filled or hollow circle in a band; inner cells are split into two halves with a short vertical stroke and a circle on that stroke. Some cells are blank. Deduce the governing pattern and determine how every missing cell must look. The completed grid must satisfy the same logic everywhere. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the fully completed grid.
Video
Data video source
Video prompt
A 5×5 grid shows symbolic cells following a fixed rule: the two outer columns are mirror-identical; each outer cell is split into three horizontal bands with one filled or hollow circle in a band; inner cells are split into two halves with a short vertical stroke and a circle on that stroke. Some cells are blank. Deduce the governing pattern and determine how every missing cell must look. The completed grid must satisfy the same logic everywhere. Only modify what the task requires; keep the background and any other element unchanged.
Bridges
Image sequence
SFT-Image source
Image prompt
Hashiwokakero puzzle: circles with numbers are islands, gray dots are empty grid positions. Draw straight horizontal or vertical bridges between islands so that: each island's bridge count matches its number; no two bridges cross; at most 2 bridges between any pair; all islands are connected. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Hashiwokakero puzzle: circles with numbers are islands, gray dots are empty grid positions. Draw straight horizontal or vertical bridges between islands so that: each island's bridge count matches its number; no two bridges cross; at most 2 bridges between any pair; all islands are connected. Generate a video that adds the bridges step by step until every island's count is satisfied and all islands are connected. Only modify what the task requires; keep the background and any other element unchanged.
Maze Directional Arrow Navigation
Image sequence
SFT-Image source
Image prompt
From the red start arrow, navigate to the green goal arrow on the grid. Each move is relative to the arrow's current facing direction (forward, backward, left, or right), and the arrow turns to face each move. Output a single image showing the complete route as a blue line connecting the red start arrow to the green goal arrow. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
Animate the arrow from the red initial position to the green final position. Each command is relative to the arrow's current orientation: forward, backward, left, or right. After moving, update the arrow so it points in the movement direction. Only modify what the task requires; keep the background and any other element unchanged.
Grid Avoid Red Block
Image sequence
SFT-Image source
Image prompt
The scene shows a 10x10 grid with a blue start square (containing a yellow circular agent), a green end square, and multiple red filled rectangles indicating obstacles. Starting from the blue start square, the agent can move to adjacent cells (up, down, left, right). The goal is to move the agent to the green end square along the shortest path without entering any cells containing red filled rectangles. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the complete solution path drawn as a cyan line.
Video
Data video source
Video prompt
The scene shows a 10x10 grid with a blue start square (containing a yellow circular agent), a green end square, and multiple red filled rectangles indicating obstacles. Starting from the blue start square, the agent can move to adjacent cells (up, down, left, right). The goal is to move the agent to the green end square along the shortest path without entering any cells containing red filled rectangles. Only modify what the task requires; keep the background and any other element unchanged.
Maze Ring Path Trace
Image sequence
SFT-Image source
Image prompt
A circular ring maze with black walls and light corridors on a black field has two red dots (one near the hub, one toward the rim). Draw a single continuous red stroke that connects the inner dot to the outer dot along the corridor midlines. The red path must never touch or overlap black wall pixels, must not self-overlap, and should match the red of the dots. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the ring maze with the completed red path drawn.
Video
Data video source
Video prompt
A circular ring maze: black walls and light corridors on a black field, with two red dots (one near the hub, one toward the rim). Animate a single continuous red stroke that grows along the corridor midlines, connecting the inner dot to the outer dot. The red path must never touch or overlap black wall pixels, must not self-overlap, and should match the red of the dots. Hold the start layout briefly, reveal the path smoothly, then hold the completed route. Only modify what the task requires; keep the background and any other element unchanged.
Track Object Movement
Image sequence
Data image source
Image prompt
The object marked with a green border is the only object that moves. It moves horizontally to align directly below the object with a red star at its center. Track the movement with the green border as the object moves. Only modify what the task requires; keep the background and any other element unchanged. Output 2 images showing uniform movement progress.
Video
Data video source
Video prompt
The object marked with a green border is the only object that moves. It moves horizontally to align directly below the object with a red star at its center. Track the movement with the green border as the object moves. Only modify what the task requires; keep the background and any other element unchanged.
Outline Innermost Square
Image sequence
Data image source
Image prompt
The scene shows multiple concentric squares (same center, from largest to smallest). Identify the innermost square and outline it with a blue square outline. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The scene shows multiple concentric squares (same center, from largest to smallest). Identify the innermost square and outline it with a blue square outline. Show the solution step by step. Only modify what the task requires; keep the background and any other element unchanged.
3D Views Perspective
Image sequence
SFT-Image source
Image prompt
The blue faces of the small cubes are the ones facing world +Y (the top side). A red arrow points vertically downward on the screen, with its head landing on the centroid of one of the blue top faces. Options A/B/C below show top-down (XZ) views with rotations of 0/90/180/270 degrees, and each option mirrors the X axis to match a left-right intuition of the top-down view. Please choose the option whose top-down projection of the blue faces exactly matches the main view. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image: the same 3D view as shown, with the matching option circled.
Video
Data video source
Video prompt
The blue faces of the small cubes are the ones facing world +Y (the top side). A red arrow points vertically downward on the screen, with its head landing on the centroid of one of the blue top faces. Options A/B/C below show top-down (XZ) views with rotations of 0/90/180/270 degrees, and each option mirrors the X axis to match a left-right intuition of the top-down view. The video marks the option whose top-down projection of the blue faces matches the main view. Only modify what the task requires; keep the background and any other element unchanged.
Paper Fold Dotted Line
Image sequence
SFT-Image source
Image prompt
Fold the paper along the dotted line, folding the portion to the right of the line over onto the left, keeping the paper appearance unchanged except for the fold. Output 4 images showing the fold progressing from flat to fully folded at uniform stages.
Video
Data video source
Video prompt
Fold the paper along the dotted line, folding the portion to the right of the line over onto the left, with fixed camera and unchanged paper appearance except the fold motion.
Count 3D Blocks
Image sequence
SFT-Image source
Image prompt
A 3D structure built from unit cubes is shown. Count the total number of unit cubes, including any hidden or occluded ones. Separate the structure into its horizontal layers from bottom to top and count the cubes one by one; the 'Number:' counter in the top-right shows the final total. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final state with every cube counted and the total displayed.
Video
Data video source
Video prompt
Count every unit cube in the 3D block structure, including blocks that are hidden behind others. Use a static viewpoint: explode the structure into horizontal layers (bottom to top), then go through the cubes in order, turning each one blue as it is counted (previously counted cubes remain blue). A large 'Number:' counter in the upper-right rises one by one to the final total. The layers stay separated at the end. Only modify what the task requires; keep the background and any other element unchanged.
Symmetry Random
Image sequence
SFT-Image source
Image prompt
The scene shows a grid with a random scatter pattern on the left half. Expand the scatter pattern symmetrically to the right half by mirroring it across the vertical axis, creating a complete symmetric pattern. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
The scene shows a grid with a random scatter pattern on the left half. Expand the scatter pattern symmetrically to the right half by mirroring it across the vertical axis, creating a complete symmetric pattern. Only modify what the task requires; keep the background and any other element unchanged.
Maze Circular Route To Center
Image sequence
SFT-Image source
Image prompt
In the circular maze, move from the green start to the red goal along a shortest valid route, and draw a continuous blue line. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the maze with the completed route drawn.
Video
Data video source
Video prompt
In the circular maze, move from the green start to the red goal along a shortest valid route, and draw a continuous blue line. Only modify what the task requires; keep the background and any other element unchanged.
Directed Graph Navigation
Image sequence
Data image source
Image prompt
The scene shows a network of nodes connected by directed edges (edges with arrows indicating direction) with a green starting node, a red ending node, and a blue triangular agent positioned at the green starting node. The agent can only move along edges in the direction they point (from the source node to the target node, cannot move backwards), moving from one node to an adjacent node each step. Move the blue triangular agent from the green starting node to the red ending node along the path with the minimum number of steps. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the solution path with traversed edges highlighted in yellow.
Video
Data video source
Video prompt
The scene shows a network of nodes connected by directed edges (edges with arrows indicating direction) with a green starting node, a red ending node, and a blue triangular agent positioned at the green starting node. The agent can only move along edges in the direction they point (from the source node to the target node, cannot move backwards), moving from one node to an adjacent node each step. Move the blue triangular agent from the green starting node to the red ending node along the path with the minimum number of steps. Only modify what the task requires; keep the background and any other element unchanged.
Bfs
Image sequence
SFT-Image source
Image prompt
The scene shows a tree structure with nodes connected by edges, a green starting node at the top, a red ending node at the bottom, and a yellow smiley agent positioned at the green starting node. The agent can move along edges, visiting one node per step. Move the yellow smiley agent from the green starting node to the red ending node along the unique tree path. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the solution path with traversed edges highlighted in yellow.
Video
Data video source
Video prompt
The scene shows a tree structure with nodes connected by edges, a green starting node at the top, a red ending node at the bottom, and a yellow smiley agent positioned at the green starting node. The agent can move along edges, visiting one node per step. Move the yellow smiley agent from the green starting node to the red ending node along the unique tree path, showing the complete movement process step by step. Only modify what the task requires; keep the background and any other element unchanged.
Maze Shortest Path Start To Goal
Image sequence
SFT-Image source
Image prompt
Move from the green start dot to the red goal dot along a shortest valid path, using only up/down/left/right moves, and draw a blue line tracing the path. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the maze with the completed path drawn.
Video
Data video source
Video prompt
Move from the green start dot to the red goal dot along a shortest valid path, using only up/down/left/right moves, and draw a blue line as you move. Only modify what the task requires; keep the background and any other element unchanged.
Grid Color Sequence
Image sequence
SFT-Image source
Image prompt
The scene shows a 10x10 grid with a green start point, a red end point, and colored cells (orange, yellow, and blue). A purple circular agent is positioned at the green start point. The agent can move to adjacent cells (up, down, left, right). Starting from the green start point, the agent must visit the colored cells in order (orange, then yellow, then blue), taking the shortest path between each consecutive pair of colored cells. The agent must not pass through any colored cell whose turn to be visited has not yet come. The agent is allowed to pass through the red end point when visiting the colored cells if needed. After visiting all colored cells in sequence, the agent must reach the red end point, also following the shortest path. Only modify what the task requires; keep the background and any other element unchanged. Output one image each time the agent reaches the next colored cell, with the path segment from the previous position drawn as a cyan line.
Video
Data video source
Video prompt
The scene shows a 10x10 grid with a green start point, a red end point, and colored cells (orange, yellow, and blue). A purple circular agent is positioned at the green start point. The agent can move to adjacent cells (up, down, left, right). Starting from the green start point, the agent must visit the colored cells in order (orange, then yellow, then blue), taking the shortest path between each consecutive pair of colored cells. The agent must not pass through any colored cell whose turn to be visited has not yet come. The agent is allowed to pass through the red end point when visiting the colored cells if needed. After visiting all colored cells in sequence, the agent must reach the red end point, also following the shortest path. Only modify what the task requires; keep the background and any other element unchanged.
Maze Dual Dot Pathfinding
Image sequence
SFT-Image source
Image prompt
In the hex maze, move from the green start cell to the red end cell along a shortest valid path, and draw a continuous blue line. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the maze with the completed path drawn.
Video
Data video source
Video prompt
In the hex maze, move from the green start cell to the red end cell along a shortest valid path, and draw a continuous blue line. Only modify what the task requires; keep the background and any other element unchanged.
Select Leftmost Shape
Image sequence
Data image source
Image prompt
Several colored shapes appear on a white background. Draw a red circle around the leftmost shape only. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image of the final result showing the leftmost shape marked with the red circle and every other shape unchanged.
Video
Data video source
Video prompt
Several colored shapes appear on a white background. Draw a red circle around the leftmost shape only. Only modify what the task requires; keep the background and any other element unchanged.
Maze Elf To Gift Path
Image sequence
SFT-Image source
Image prompt
Move the elf to the gift through valid cells, avoid all icy lake tiles, and draw a green path line tracing the route. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the maze with the completed path drawn.
Video
Data video source
Video prompt
Move the elf to the gift through valid cells, avoid all icy lake tiles, and draw a green path line while moving. Only modify what the task requires; keep the background and any other element unchanged.
Symmetry Shape
Image sequence
SFT-Image source
Image prompt
[Scenario] The image shows an N×N grid with a contiguous shape made of coloured blocks confined to the left half. A dashed vertical line marks the centre axis of mirror symmetry.
[Task] Fill the right half by mirroring every shape block across the centre axis, yielding a complete vertically-symmetric figure. The grid is fixed; the camera does not move. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
[Scenario] The image shows an N×N grid with a contiguous shape made of coloured blocks confined to the left half. A dashed vertical line marks the centre axis of mirror symmetry.
[Task] Produce a video that fills the right half by mirroring every shape block across the centre axis, yielding a complete vertically-symmetric figure. The grid is fixed; the camera does not move. Only modify what the task requires; keep the background and any other element unchanged.
Grid Obtaining Award
Image sequence
SFT-Image source
Image prompt
The scene shows a 10x10 grid with a green start cell, a red end cell, and 4 diamond reward items scattered across it. A circular agent starts at the green cell and can move to adjacent cells (up, down, left, right). The agent collects a reward by moving to its cell, and once collected the reward disappears. Follow the shortest path which collects all 4 diamond rewards before arriving at the red end cell. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each reward is collected, plus a final image when the agent arrives at the red end cell, each with the path from the previous position drawn as a cyan line.
Video
Data video source
Video prompt
The scene shows a 10x10 grid with a green start cell, a red end cell, and 4 diamond reward items scattered across it. A circular agent starts at the green cell and can move to adjacent cells (up, down, left, right). The agent collects a reward by moving to its cell, and once collected the reward disappears. Produce a video that follows the shortest path which collects all 4 diamond rewards before arriving at the red end cell. Only modify what the task requires; keep the background and any other element unchanged.
Maze
Image sequence
Data image source
Image prompt
The scene shows a 15×15 grid maze with dark walls and white pathways. A green circular marker indicates the starting position, and a red flag marks the end position. Starting from the green start position, navigate through the maze by moving along the white pathways. You can move to adjacent cells (up, down, left, right) but cannot pass through the dark walls. The goal is to find and demonstrate the complete path from the green start to the red flag end position, covering every step of the journey through the maze. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the complete path drawn as a yellow line from the green start to the red flag end position.
Video
Data video source
Video prompt
The scene shows a 15×15 grid maze with dark walls and white pathways. A green circular marker indicates the starting position, and a red flag marks the end position. Starting from the green start position, navigate through the maze by moving along the white pathways. You can move to adjacent cells (up, down, left, right) but cannot pass through the dark walls. The goal is to find and demonstrate the complete path from the green start to the red flag end position, showing each step of the journey through the maze. Only modify what the task requires; keep the background and any other element unchanged.
Find Keys And Open Doors
Image sequence
SFT-Image source
Image prompt
[Scenario] The image shows a maze with black walls and white paths. The green circle is the agent's starting position. A coloured diamond marks a key, and a coloured hollow square marks a door.
[Rules]
1. The agent must first walk through the maze to the key cell to collect it.
2. After the key is collected, the agent must walk to the door cell.
3. The agent may only move along white path cells; it cannot cross black walls.
[Task] The agent collects the key and then reaches the door. The camera is fixed; the maze layout does not change. Only modify what the task requires; keep the background and any other element unchanged. Output two images: the first showing the path from start to the key drawn as a yellow line, the second showing the path from the key to the door drawn as a yellow line.
Video
Data video source
Video prompt
[Scenario] The image shows a maze with black walls and white paths. The green circle is the agent's starting position. A coloured diamond marks a key, and a coloured hollow square marks a door.
[Rules]
1. The agent must first walk through the maze to the key cell to collect it.
2. After the key is collected, the agent must walk to the door cell.
3. The agent may only move along white path cells; it cannot cross black walls.
[Task] Produce a video that animates the agent collecting the key and then reaching the door. The camera is fixed; the maze layout does not change. Only modify what the task requires; keep the background and any other element unchanged.
Lego Construction Assembly
Image sequence
Data image source
Image prompt
This is LEGO assembly step 2. The scene shows a partial model on the right and a 1x1 orange brick on the left in a callout box. Take the orange brick and attach it to the model at the position indicated by the red arrow. Move the brick smoothly from the callout to its destination, align it correctly, and snap it into place. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
This is LEGO assembly step 2. The scene shows a partial model on the right and a 1x1 orange brick on the left in a callout box. Take the orange brick and attach it to the model at the position indicated by the red arrow. Move the brick smoothly from the callout to its destination, align it correctly, and snap it into place. Only modify what the task requires; keep the background and any other element unchanged.
Grid Go Through Block
Image sequence
SFT-Image source
Image prompt
The scene shows a 10×10 grid with a green start cell, a red end cell, and colored block cells (blue, brown, yellow). An orange circular agent starts at the green cell and can move to adjacent cells (up, down, left, right). The agent must visit all colored blocks (blue, brown, yellow) in order, taking the shortest path between consecutive targets, then reach the red end cell. Only modify what the task requires; keep the background and any other element unchanged. Output one image each time the agent reaches the next block, plus a final image when the agent arrives at the red end cell, each with the path segment from the previous position drawn as a cyan line.
Video
Data video source
Video prompt
The scene shows a 10×10 grid with a green start cell, a red end cell, and colored block cells (blue, brown, yellow). An orange circular agent starts at the green cell and can move to adjacent cells (up, down, left, right). The agent must visit all colored blocks (blue, brown, yellow) in order, taking the shortest path between consecutive targets, then reach the red end cell. Only modify what the task requires; keep the background and any other element unchanged.
Horizontal Symmetry Reflection
Image sequence
SFT-Image source
Image prompt
Reflect the colored pattern across the central horizontal axis. Keep the existing pattern unchanged, and fill in its mirror image on the opposite side so the figure becomes symmetric. The central horizontal symmetry axis is highlighted as a red line. Output 1 image showing the final symmetric figure.
Video
Data video source
Video prompt
Build the horizontal-axis mirror of the pattern. Do not modify the original colored blocks; add each reflected cell across the red center line step by step, following top-to-bottom, left-to-right order, until the figure is symmetric. Keep the camera fixed with no scaling or panning.
Maze Glowing Path Trace
Image sequence
SFT-Image source
Image prompt
Find the correct route through the maze's open corridors from the blue point at the top to the red point at the bottom, passing through the green numbered marks on the way. Draw a clear blue path line along the route, keeping the maze and all walls unchanged. Output 1 image showing the maze with the completed path drawn.
Video
Data video source
Video prompt
Animate a bright path tracing from the blue point at the top through the maze's open corridors toward the red point at the bottom, highlighting each green numbered mark it passes. Draw a clear blue path line along the route as the animation progresses. Keep the maze and all walls fixed while the glowing path moves smoothly through the correct route. Static shot.
3D Cube Unfold Net
Image sequence
SFT-Image source
Image prompt
A 1024×1024 reasoning sheet shows a six-square cube net in the center—each face has a distinct fill color so faces are identifiable when folded. Four isometric cubes labeled A–D sit in the corners; each shows three visible faces in matching colors. Treat the central cross as an unfolded cube and decide which single corner orientation is consistent with the net's adjacency, so the face colors line up with their neighbors. Mark the answer by drawing a thick green rectangular frame around the correct option. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final sheet with a thick green frame around the correct option.
Video
Data video source
Video prompt
A 1024×1024 reasoning sheet shows a six-square cube net in the center—each face has a distinct fill color so faces are identifiable when folded. Four isometric cubes labeled A–D sit in the corners; each shows three visible faces in matching colors. Exactly one cube orientation is consistent with the net’s adjacency. The ground-truth video holds the puzzle, then draws a thick green rectangular frame around the correct option. Only modify what the task requires; keep the background and any other element unchanged.
Key Door Matching
Image sequence
SFT-Image source
Image prompt
[Scenario] The image shows a maze with black walls and white paths. The green circle is the agent's starting position. Several coloured diamonds mark keys and the corresponding hollow coloured squares mark doors. Each key matches the door of the same colour.
[Rules]
1. The agent must first walk through the maze to the Blue key cell to collect it (the key then disappears).
2. After collecting the Blue key, the agent must walk to the Blue door cell.
3. The agent may only move along white path cells; it cannot cross black walls.
[Task] The agent collects the Blue key and then reaches the matching Blue door. The camera is fixed; the maze layout, keys, and doors do not change. Only modify what the task requires; keep the background and any other element unchanged. Output two images: the first showing the path from start to the key drawn as a yellow line, the second showing the path from the key to the door drawn as a yellow line.
Video
Data video source
Video prompt
[Scenario] The image shows a maze with black walls and white paths. The green circle is the agent's starting position. Several coloured diamonds mark keys and the corresponding hollow coloured squares mark doors. Each key matches the door of the same colour.
[Rules]
1. The agent must first walk through the maze to the Blue key cell to collect it (the key then disappears).
2. After collecting the Blue key, the agent must walk to the Blue door cell.
3. The agent may only move along white path cells; it cannot cross black walls.
[Task] Produce a video that animates the agent collecting the Blue key and then reaching the matching Blue door. The camera is fixed; the maze layout, keys, and doors do not change. Only modify what the task requires; keep the background and any other element unchanged.
Construction Blueprint
Image sequence
SFT-Image source
Image prompt
In the scene, the upper structure has a missing piece outlined with a dashed line. There are 4 candidate pieces below. Sequentially check each candidate from left to right: highlight the current candidate being examined with a frame, preview how the piece fits in the gap, then judge it: if the shape matches, turn its box green and mark it with a green ✓; if it doesn't match, turn its box red and mark it with a red ✗. Then move to the next candidate. Once the matching piece is found, output 1 image showing it moved into the gap to complete the structure. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
In the scene, the upper structure has a missing piece outlined with a dashed line. There are 4 candidate pieces below. The video sequentially checks each candidate from left to right: highlights the current candidate being examined with a frame, previews how the piece fits in the gap, then judges it: if the shape matches, its box turns green and is marked with a green ✓; if it doesn't match, its box turns red and is marked with a red ✗. It then moves to the next candidate. Once the matching piece is found, an animation demonstrates it moving into the gap to complete the structure. Only modify what the task requires; keep the background and any other element unchanged.
Undirected Graph Navigation
Image sequence
SFT-Image source
Image prompt
The scene shows a network of nodes connected by undirected edges (edges without arrows) with a green starting node, a red ending node, and a purple triangular agent positioned at the green starting node. The agent can move along any edge, traversing it from one end to the other in either direction, moving from one node to an adjacent node each step. Move the purple triangular agent from the green starting node to the red ending node along the path with the minimum number of steps. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the solution path with traversed edges highlighted in yellow.
Video
Data video source
Video prompt
The scene shows a network of nodes connected by undirected edges (edges without arrows) with a green starting node, a red ending node, and a purple triangular agent positioned at the green starting node. The agent can move along any edge, traversing it from one end to the other in either direction, moving from one node to an adjacent node each step. Move the purple triangular agent from the green starting node to the red ending node along the path with the minimum number of steps. Only modify what the task requires; keep the background and any other element unchanged.
Maze Stepwise Arrow Route
Image sequence
SFT-Image source
Image prompt
Move from the red start dot to the red letter target cell 'L' along a shortest valid path, and draw a blue line tracing the path. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the maze with the completed path drawn.
Video
Data video source
Video prompt
Move from the red start dot to the red letter target cell L along a shortest valid path, and draw a blue line step by step. Only modify what the task requires; keep the background and any other element unchanged.
Cube Net Folding
Image sequence
SFT-Image source
Image prompt
Fold an unfolded cube net along its edges into a complete cube: keep one face as the stationary base while the other five fold up onto it, so the colored, lettered faces end up on the outside of the cube. Keep the fold edges clearly visible. Output 4 images showing the fold progressing from the flat net to the complete cube at uniform stages.
Video
Data video source
Video prompt
Animate folding a cube net along its edges into a solid cube. Keep one face as the stationary base while the other five fold up onto it, so the colored, lettered faces end up on the outside. Display crisp fold lines during the motion. Do not change anything except the folding transformation.
Rotation
Image sequence
Data image source
Image prompt
A 6-block sculpture sits fixed on a table. First frame: Your camera is tilted at 36° elevation, viewing from 10° azimuth. Final frame: Your camera remains at 36° elevation, but rotates horizontally to 190° azimuth. This is a 180-degree rotation. The camera rotates horizontally around the sculpture; try to maintain the tilted viewing angle throughout. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the camera's horizontal rotation at uniform progress intervals.
Video
Data video source
Video prompt
A 6-block sculpture sits fixed on a table. First frame: Your camera is tilted at 36° elevation, viewing from 10° azimuth. Final frame: Your camera remains at 36° elevation, but rotates horizontally to 190° azimuth. This is a 180-degree rotation Create a smooth video showing the camera's horizontal rotation around the sculpture, and try to maintain the tilted viewing angle throughout. Only modify what the task requires; keep the background and any other element unchanged.
Grid Shortest Path
Image sequence
SFT-Image source
Image prompt
The scene shows a 10x10 grid with a red start square (containing a orange circular agent) and a pink end square. Starting from the red start square, the agent can move to adjacent cells (up, down, left, right). The goal is to move the agent to the pink end square along the shortest path. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the complete shortest path drawn as a cyan line, with the agent at the end position.
Video
Data video source
Video prompt
The scene shows a 10x10 grid with a red start square (containing a orange circular agent) and a pink end square. Starting from the red start square, the agent can move to adjacent cells (up, down, left, right). The goal is to move the agent to the pink end square along the shortest path. Only modify what the task requires; keep the background and any other element unchanged.
Arc Continuation Matching
Image sequence
SFT-Image source
Image prompt
One smooth black arc enters from the left and meets the left edge of a vertical light-gray mask in the center. On the right side of the mask, five labeled segments A through E start from the mask edge. Exactly one of them continues the same hidden arc through the band. Determine the true continuation, with the masked band removed so the continuation is visible, and the correct labeled segment emphasized (for example with a red stroke or glow). Output 1 image showing the final state with the correct continuation segment marked.
Video
Data video source
Video prompt
Spatial puzzle: continue the left arc mentally through the masked strip and pick which right-hand label (A–E) matches. Initial frame shows the band; final frame removes it to confirm. Static camera, square frame, no zoom. Animate only by erasing the mask quickly; arcs stay fixed. End by visually highlighting the matching segment only.
Symmetry Completion
Image sequence
Data image source
Image prompt
Complete this striped pattern by filling in the missing cells on the right side. The left half shows horizontal stripes where entire rows are either filled or empty. Mirror the left half across the vertical center line to complete the symmetric striped pattern. Keep the camera view fixed in the top-down perspective and maintain all existing cells unchanged. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the completed symmetric pattern.
Video
Data video source
Video prompt
Complete this striped pattern by filling in the missing cells on the right side. The left half shows horizontal stripes where entire rows are either filled or empty. Mirror the left half across the vertical center line to complete the symmetric striped pattern. Keep the camera view fixed in the top-down perspective and maintain all existing cells unchanged. Stop the video when the pattern is complete. Only modify what the task requires; keep the background and any other element unchanged.
Maze Marker Draw Start To End
Image sequence
SFT-Image source
Image prompt
Move from the green start marker to the red end marker along a shortest valid path, and draw a continuous blue line on the route. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the maze with the completed path drawn.
Video
Data video source
Video prompt
Move from the green start marker to the red end marker along a shortest valid path, and keep a continuous blue line on the route. Only modify what the task requires; keep the background and any other element unchanged.
Locate Topmost Unobscured Figure
Image sequence
Data image source
Image prompt
Multiple shapes partially overlap. Outline the topmost (unobscured) shape with a red outline. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
Multiple shapes partially overlap. Outline the topmost (unobscured) shape with a red outline. Only modify what the task requires; keep the background and any other element unchanged.
Multiple Keys For One Door
Image sequence
Data image source
Image prompt
[Scenario] The image shows a maze with black walls and white paths. The green circle is the agent's starting position. Multiple coloured diamonds mark keys (each a different colour) and a single coloured hollow square marks the door.
[Rules]
1. The agent must collect EVERY key (the key disappears the moment it is picked up).
2. After collecting all keys, the agent must walk to the door cell.
3. The agent may only move along white path cells; it cannot cross black walls.
4. Pick the key-collection order that minimises the total travel distance (agent → first key → … → last key → door).
[Task] The agent collects all keys (in the optimal order) and then reaches the door. The camera is fixed; the maze layout does not change. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each key is collected and one after reaching the door, with each path segment drawn as a yellow line.
Video
Data video source
Video prompt
[Scenario] The image shows a maze with black walls and white paths. The green circle is the agent's starting position. Multiple coloured diamonds mark keys (each a different colour) and a single coloured hollow square marks the door.
[Rules]
1. The agent must collect EVERY key (the key disappears the moment it is picked up).
2. After collecting all keys, the agent must walk to the door cell.
3. The agent may only move along white path cells; it cannot cross black walls.
4. Pick the key-collection order that minimises the total travel distance (agent → first key → … → last key → door).
[Task] Produce a video that animates the agent collecting all keys (in the optimal order) and then reaching the door. The camera is fixed; the maze layout does not change. Only modify what the task requires; keep the background and any other element unchanged.
Maze Triangle Pickup And Place
Image sequence
SFT-Image source
Image prompt
Move the red triangle to pick up the white printer, then carry it to the highlighted target area and place it there. Draw an orange line tracing the movement path. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final state with the completed path drawn.
Video
Data video source
Video prompt
Move the red triangle to pick up the white printer, then carry it to the highlighted target area and place it there. Draw an orange line to trace the movement path. Only modify what the task requires; keep the background and any other element unchanged.
Maze Portrait Red Path
Image sequence
SFT-Image source
Image prompt
In this top-down maze, a green square marks the start at the top-left and a blue square marks the goal at the bottom-right. Black is walls, white is corridors. Draw a continuous red route that connects the two markers, stays inside the white corridors, remains centered away from black walls, and does not cut through walls. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the maze with the completed red path drawn.
Video
Data video source
Video prompt
Connect the green start marker to the blue goal marker with a single red path through the white passages only. The maze is orthogonal, portrait-shaped on a square frame. Keep the stroke in the middle of corridors so it never touches black. Static viewpoint; show the full solution path from start to goal. Only modify what the task requires; keep the background and any other element unchanged.
Grid Avoid Obstacles
Image sequence
SFT-Image source
Image prompt
The scene shows a 10×10 grid with a blue start square (containing a yellow circular agent), a red end square, and several black 'X' markers indicating obstacles. Starting from the blue start square, the agent can move to adjacent cells (up, down, left, right). The goal is to move the agent to the red end square along the shortest path without entering any cell containing a black 'X' obstacle. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the complete solution path drawn as a cyan line.
Video
Data video source
Video prompt
The scene shows a 10×10 grid with a blue start square (containing a yellow circular agent), a red end square, and several black 'X' markers indicating obstacles. Starting from the blue start square, the agent can move to adjacent cells (up, down, left, right). The goal is to move the agent to the red end square along the shortest path without entering any cell containing a black 'X' obstacle. Only modify what the task requires; keep the background and any other element unchanged.
Grid Highest Cost
Image sequence
SFT-Image source
Image prompt
A 4x4 grid shows integer costs in each cell. The top-left cell is green (start) and the bottom-right cell is red (goal). A yellow Pac-Man starts on the green cell. Find a path from start to goal that maximizes the sum of costs collected (each time Pac-Man enters a cell, add that cell's cost). Only orthogonal moves (up/down/left/right). Keep the grid lines and cell cost numbers unchanged; only Pac-Man should move. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the complete highest-cost path drawn as an orange line from the green start cell to the red goal cell.
Video
Data video source
Video prompt
A 4x4 grid shows integer costs in each cell. The top-left cell is green (start) and the bottom-right cell is red (goal). A yellow Pac-Man starts on the green cell. Animate Pac-Man moving step-by-step along a path from start to goal that maximizes the sum of costs collected (each time Pac-Man enters a cell, add that cell's cost). Only orthogonal moves (up/down/left/right). Keep the grid lines and cell cost numbers unchanged; only Pac-Man should move. Only modify what the task requires; keep the background and any other element unchanged.
Maze Shortest Path Horizontal Vertical
Image sequence
SFT-Image source
Image prompt
Move from the green start marker to the red goal marker along a shortest valid path, using only horizontal and vertical moves, and draw a green line tracing the path. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the maze with the completed path drawn.
Video
Data video source
Video prompt
Move from the green start marker to the red goal marker along a shortest valid path, using only horizontal and vertical moves, and draw a green line as you move. Only modify what the task requires; keep the background and any other element unchanged.
Grid Number Sequence
Image sequence
SFT-Image source
Image prompt
Guide the agent from green to red, passing through yellow numbers in sequence (1→2→3→...). Move only up, down, left, or right. Optimize the path length between each pair of consecutive waypoints. Only modify what the task requires; keep the background and any other element unchanged. Output one image each time the agent reaches the next numbered cell, plus a final image when the agent arrives at the red end cell, each with the path segment from the previous position drawn as a cyan line.
Video
Data video source
Video prompt
Guide the agent from green to red, passing through yellow numbers in sequence (1→2→3→...). Move only up, down, left, or right. Optimize the path length between each pair of consecutive waypoints. Only modify what the task requires; keep the background and any other element unchanged.
Maze Cloud Route Trace
Image sequence
SFT-Image source
Image prompt
In the hex-cell maze, move from the green start marker to the red goal marker along a shortest valid path, and draw a continuous green line without crossing walls. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the maze with the completed path drawn.
Video
Data video source
Video prompt
In the hex-cell maze, move from the green start marker to the red goal marker along a shortest valid path, and draw a continuous green line without crossing walls. Only modify what the task requires; keep the background and any other element unchanged.
3D Pattern Completion
Image sequence
SFT-Image source
Image prompt
A 1024×1024 worksheet shows an incomplete blue voxel solid in the upper panel and three candidate pieces A, B, and C in the lower panel. Inspect the carved cavity of the incomplete 3D shape and determine which option matches the missing piece. Blue face tones indicate orientation and black lines indicate cube boundaries; the upper solid itself remains incomplete. Output 1 image showing the final state with the correct option highlighted.
Video
Data video source
Video prompt
A 1024×1024 worksheet shows an incomplete blue voxel solid in the upper panel and three candidate pieces A, B, and C in the lower panel. The upper solid does not get completed during the video. Instead, the same incomplete 3D shape rotates through a full 360-degree turn so its carved cavity can be inspected from all sides. Use that rotation to determine which option matches the missing piece. Blue face tones indicate orientation, black lines indicate cube boundaries, and the final frame may highlight the correct option, but the upper solid itself remains incomplete throughout.
Vertical Symmetry Reflection
Image sequence
SFT-Image source
Image prompt
Reflect the colored pattern across the central vertical axis. Keep the existing pattern unchanged, and fill in its mirror image on the opposite side so the figure becomes symmetric. The central vertical symmetry axis is highlighted as a red line. Output 1 image showing the final symmetric figure.
Video
Data video source
Video prompt
Build the vertical-axis mirror of the pattern. Do not modify the original colored blocks; add each reflected cell across the red center line step by step, following top-to-bottom, left-to-right order, until the figure is symmetric. Keep the camera fixed with no scaling or panning.
Move 2 Object To 2 Target
Image sequence
Data image source
Image prompt
In the scene there are two objects and their corresponding target outlines; each outline matches its object in color and shape. Move each object to its matching outline via shortest path. Both objects move together at the same time and reach their target outlines simultaneously. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the objects at uniform movement progress.
Video
Data video source
Video prompt
In the scene there are two objects and their corresponding target outlines; each outline matches its object in color and shape. Move each object to its matching outline via shortest path. Both objects move together at the same time and reach their target outlines simultaneously. Show the movement step by step. Only modify what the task requires; keep the background and any other element unchanged.
Ballcolor
Image sequence
SFT-Image source
Image prompt
In the scene, there are 4 ball clusters: red cluster A has 5 balls, and there are other clusters with different numbers of balls. The global TOTAL label shows the sum 25. Control the red cluster A to absorb other clusters. You can only absorb clusters that are smaller than the red cluster. Find the correct sequence to absorb all clusters step by step until all balls become red. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each merge event, showing the state after the red cluster absorbs a smaller cluster.
Video
Data video source
Video prompt
In the scene, there are 4 ball clusters: red cluster A has 5 balls, and there are other clusters with different numbers of balls. The global TOTAL label shows the sum 25. Control the red cluster A to absorb other clusters. You can only absorb clusters that are smaller than the red cluster. Find the correct sequence to absorb all clusters step by step until all balls become red. Only modify what the task requires; keep the background and any other element unchanged.
Object Trajectory
Image sequence
SFT-Image source
Image prompt
The scene contains a diamond object and a dashed target position (indicated by a dashed outline of the same shape). Keep the dashed target position unchanged. Move the diamond object to the dashed target position along the shortest path, ensuring it completely overlaps with the target. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the object at uniform movement progress.
Video
Data video source
Video prompt
The scene contains a diamond object and a dashed target position (indicated by a dashed outline of the same shape). Keep the dashed target position unchanged. Move the diamond object to the dashed target position along the shortest path, ensuring it completely overlaps with the target. Only modify what the task requires; keep the background and any other element unchanged.
Grid Shift
Image sequence
SFT-Image source
Image prompt
In the scene, there is a 4x4 grid with 5 yellow square blocks, each outlined in black, placed at different positions. Translate all blocks upward by exactly 2 steps, moving simultaneously and uniformly. Each block shifts 2 cells toward the top direction, and all blocks must remain completely within the grid boundaries. The goal is to achieve a configuration where every block has been moved exactly 2 steps upward to reach its final position. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each step of the movement, showing all blocks at their new positions.
Video
Data video source
Video prompt
In the scene, there is a 4x4 grid with 5 yellow square blocks, each outlined in black, placed at different positions. Translate all blocks upward by exactly 2 steps, moving simultaneously and uniformly. Each block shifts 2 cells toward the top direction, and all blocks must remain completely within the grid boundaries. The goal is to achieve a configuration where every block has been moved exactly 2 steps upward to reach its final position. Only modify what the task requires; keep the background and any other element unchanged.
Tangram Target Shape Completion
Image sequence
SFT-Image source
Image prompt
A top-down grid puzzle: the main board shows fixed numbered pieces and a hatched empty silhouette; a lower panel holds the blue piece 1. Place piece 1 — rotated in 90° steps and translated — into the hatched region so it fills the silhouette exactly without overlapping any fixed piece. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final composition with piece 1 placed in the silhouette.
Video
Data video source
Video prompt
Given the current layout and the hatched target, animate piece 1 rotating and moving from the panel into the silhouette so it fits perfectly against the fixed numbered pieces with no overlap, ending in the completed composition. Only modify what the task requires; keep the background and any other element unchanged.
Track Object Movement
Image sequence
Data image source
Image prompt
The object marked with a green border is the only object that moves. It moves horizontally to align directly below the object with a red star at its center. Track the movement with the green border as the object moves. Only modify what the task requires; keep the background and any other element unchanged. Output 2 images showing uniform movement progress.
Video
Data video source
Video prompt
The object marked with a green border is the only object that moves. It moves horizontally to align directly below the object with a red star at its center. Track the movement with the green border as the object moves. Only modify what the task requires; keep the background and any other element unchanged.
Animal Matching
Image sequence
Data image source
Image prompt
Colored animal faces are on the left side of the canvas, and dark outlines of animals are on the right side. Move each colored animal face to its matching outline via the shortest path. All animal faces move together at the same time and reach their matching outlines simultaneously. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing all animals at uniform movement progress.
Video
Data video source
Video prompt
Colored animal faces are on the left side of the canvas, and dark outlines of animals are on the right side. Move each colored animal face to its matching outline via the shortest path. All animal faces move together at the same time and reach their matching outlines simultaneously. Only modify what the task requires; keep the background and any other element unchanged.
Construction Stack
Image sequence
Data image source
Image prompt
Match the left CURRENT stacks to the TARGET on the right by moving blocks (one top block per move). Keep the TARGET stacks on the right unchanged. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each block move, showing the updated stack state.
Video
Data video source
Video prompt
Match the left CURRENT stacks to the TARGET on the right by moving blocks (one top block per move). Keep the TARGET stacks on the right unchanged. Only modify what the task requires; keep the background and any other element unchanged.
Bookshelf
Image sequence
SFT-Image source
Image prompt
In the scene, there is a bookshelf with a set of books already placed, and a few purple books waiting on the right. There are gaps between the books on the shelf. Place each purple book into a gap where its height is closest to the surrounding books. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each book is inserted, showing the updated shelf.
Video
Data video source
Video prompt
In the scene, there is a bookshelf with a set of books already placed, and a few purple books waiting on the right. There are gaps between the books on the shelf. Place each purple book into a gap where its height is closest to the surrounding books. Show the insertion process step by step. Only modify what the task requires; keep the background and any other element unchanged.
Combined Objects Spinning
Image sequence
SFT-Image source
Image prompt
The scene shows 2 objects on the left side and dashed target outlines on the right side. The dashed target outlines remain completely stationary. For each object, first rotate it in place to match the orientation of its corresponding dashed target outline, then move it horizontally to the right so that it aligns exactly with and fits within its corresponding dashed target outline. All objects rotate and translate together at the same time and reach their target outlines simultaneously. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images: the first image shows every object rotated in place to its target orientation (still at its starting position), then the next 3 images show the horizontal translation at uniform progress (1/3, 2/3, ..., 3/3) until each object overlaps its dashed target.
Video
Data video source
Video prompt
The scene shows 2 objects on the left side and dashed target outlines on the right side. The dashed target outlines remain completely stationary. For each object, first rotate it in place to match the orientation of its corresponding dashed target outline, then move it horizontally to the right so that it aligns exactly with and fits within its corresponding dashed target outline. All objects rotate and translate together at the same time and reach their target outlines simultaneously. Only modify what the task requires; keep the background and any other element unchanged.
Maintain Object Identity Different Objects
Image sequence
SFT-Image source
Image prompt
The scene shows two objects, one on the left and one on the right. Swap the positions of the left and right objects. After the swap, draw an arrow below the object that was originally on the right, pointing up at it. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the objects after swapping positions with the arrow drawn.
Video
Data video source
Video prompt
The scene shows two objects, one on the left and one on the right. Swap the positions of the left and right objects. After the swap, draw an arrow below the object that was originally on the right, pointing up at it. Only modify what the task requires; keep the background and any other element unchanged.
Rotation Puzzle
Image sequence
SFT-Image source
Image prompt
Solve this pipe puzzle. The scene shows four square tiles, each containing a pipe segment. The tiles stay fixed in place and do not rotate; only the pipe inside each tile rotates. Rotate the pipe within each tile so that all pipe segments link up into a single continuous path. All pipes rotate at the same time and reach their final orientations together. Keep the camera view fixed in the top-down perspective and keep every tile in its original position. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images at uniform intervals showing the rotation progress, ending with the solved puzzle.
Video
Data video source
Video prompt
Solve this pipe puzzle. The scene shows four square tiles, each containing a pipe segment. The tiles stay fixed in place and do not rotate; only the pipe inside each tile rotates. Rotate the pipe within each tile so that all pipe segments link up into a single continuous path. All pipes rotate at the same time and reach their final orientations together. Keep the camera view fixed in the top-down perspective and keep every tile in its original position. Stop the video when all pipes are connected and the puzzle is solved. Only modify what the task requires; keep the background and any other element unchanged.
Reorder Objects
Image sequence
SFT-Image source
Image prompt
The scene contains multiple objects arranged in a horizontal line. Keep all other objects unchanged. Swap the positions of the 2nd and 3rd objects from the left using shortest paths. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
The scene contains multiple objects arranged in a horizontal line. Keep all other objects unchanged. Swap the positions of the 2nd and 3rd objects from the left using shortest paths. Only modify what the task requires; keep the background and any other element unchanged.
2D Geometric Transformation
Image sequence
Data image source
Image prompt
The scene shows a colored 2D polygon, a rotation center marked by a small circular marker, and a dashed target outline indicating the final orientation. Only orientation changes. Rotate the polygon in the clockwise direction. First note the polygon’s initial orientation and the marked center, then read the dashed outline to know the target orientation. Rotate the polygon clockwise around the center until it completely overlaps the dashed outline with no offset. Keep size unchanged, keep a line to the center, and end with the polygon exactly matching the dashed outline. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images at uniform rotation progress (1/4, 2/4, ..., 4/4).
Video
Data video source
Video prompt
The scene shows a colored 2D polygon, a rotation center marked by a small circular marker, and a dashed target outline indicating the final orientation. Only orientation changes. Rotate the polygon in the clockwise direction. First note the polygon’s initial orientation and the marked center, then read the dashed outline to know the target orientation. Rotate the polygon clockwise around the center until it completely overlaps the dashed outline with no offset. Keep size unchanged, keep a line to the center, and end with the polygon exactly matching the dashed outline. Only modify what the task requires; keep the background and any other element unchanged.
Handle Object Reappearance
Image sequence
SFT-Image source
Image prompt
There is an object in the center. Move the object right off-screen, then return along the same path to the center. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images: moved halfway right, completely disappeared off-screen, returning halfway, and back at center.
Video
Data video source
Video prompt
There is an object in the center. Move the object right off-screen, then return along the same path to the center. Only modify what the task requires; keep the background and any other element unchanged.
Multi Piece Shape Fit
Image sequence
SFT-Image source
Image prompt
Put every colored piece into its uniquely matching hole on the sorter panel. Only modify what the task requires; keep the background and any other element unchanged. Output 3 images, one after each piece is placed: image i shows the first i pieces seated in their holes.
Video
Data video source
Video prompt
Put every colored piece into its uniquely matching hole on the sorter panel. Only modify what the task requires; keep the background and any other element unchanged.
Rolling Ball
Image sequence
SFT-Image source
Image prompt
The scene shows a ball and a series of platforms arranged along a curved path. The ball rolls along the trajectory path, smoothly transitioning from one platform to the next, landing on each platform in sequence, and coming to rest on the final platform. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the ball's rolling progress at uniform intervals along the trajectory, from start to the final platform.
Video
Data video source
Video prompt
The scene shows a ball and a series of platforms arranged along a curved path. Animate the ball rolling along the trajectory path, smoothly transitioning from one platform to the next, landing on each platform in sequence, and coming to rest on the final platform. Only modify what the task requires; keep the background and any other element unchanged.
Multi Object Placement
Image sequence
SFT-Image source
Image prompt
The scene contains multiple colored objects and star markers. Keep all star markers unchanged in position. Move each colored object to the star marker with the same color using straight paths, aligning the center of each object with the center of its matching star marker. All objects move simultaneously and arrive at their matching markers at the same time. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing all objects at uniform movement progress.
Video
Data video source
Video prompt
The scene contains multiple colored objects and star markers. Keep all star markers unchanged in position. Move each colored object to the star marker with the same color using straight paths, aligning the center of each object with the center of its matching star marker. All objects move simultaneously and arrive at their matching markers at the same time. Only modify what the task requires; keep the background and any other element unchanged.
Hanoi
Image sequence
SFT-Image source
Image prompt
The image shows a Tower of Hanoi puzzle with 3 disks on 3 pegs (labelled 1–3 from left to right). Move all disks onto peg 3 so they end largest-on-bottom to smallest-on-top, following the rules: only the top disk of a peg may move, one disk at a time, and a larger disk may never rest on a smaller one. Output 5 images showing the tower after each move, in order. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
The image shows a Tower of Hanoi puzzle with 3 disks on 3 pegs (labelled 1–3 from left to right). Move all disks onto peg 3 so they end largest-on-bottom to smallest-on-top, following the rules: only the top disk of a peg may move, one disk at a time, and a larger disk may never rest on a smaller one. Generate a video that performs the optimal sequence of moves one at a time until every disk is stacked on peg 3. Only modify what the task requires; keep the background and any other element unchanged.
Object Packing
Image sequence
SFT-Image source
Image prompt
The scene shows objects on the left side and a container on the right side. Place the objects into the container one by one in the color order: purple - red - brown - cyan - orange. Each object must be placed individually in the exact order specified, and all objects must end up inside the container. Only modify what the task requires; keep the background and any other element unchanged. Output two images per object: one showing the object mid-move, and one after it is placed in the container.
Video
Data video source
Video prompt
The scene shows objects on the left side and a container on the right side. Place the objects into the container one by one in the color order: purple - red - brown - cyan - orange. Each object must be placed individually in the exact order specified, and all objects must end up inside the container. Only modify what the task requires; keep the background and any other element unchanged.
Symbol Reordering
Image sequence
SFT-Image source
Image prompt
The scene shows a sequence of 3 letter symbols arranged from left to right. Position labels (0 to 2) are shown below each symbol, indicating the target position where each symbol should move to. In symbol reordering tasks, each symbol must move from its current position to its target position indicated by the position label shown below it. The reordering rule is: first identify each symbol's current position in the sequence (from left to right, starting at position 0), then read the target position label below the symbol to determine where it should move, then move the symbol horizontally to that target position. Observe the initial sequence and read the position labels to determine where each symbol should move, ensuring all symbols reach their exact final locations as specified by the position labels. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
The scene shows a sequence of 3 letter symbols arranged from left to right. Position labels (0 to 2) are shown below each symbol, indicating the target position where each symbol should move to. In symbol reordering tasks, each symbol must move from its current position to its target position indicated by the position label shown below it. The reordering rule is: first identify each symbol's current position in the sequence (from left to right, starting at position 0), then read the target position label below the symbol to determine where it should move, then move the symbol horizontally to that target position. Observe the initial sequence and read the position labels to determine where each symbol should move. Animate each symbol moving smoothly from its current position to its target position, ensuring all symbols reach their exact final locations as specified by the position labels. Only modify what the task requires; keep the background and any other element unchanged.
Mark Tangent Point After Motion
Image sequence
SFT-Image source
Image prompt
The scene shows two circles that move only horizontally toward each other (left-to-right and right-to-left) until they touch and stop. First locate the single tangent point where the two circle boundaries meet, then draw a red circle around that tangent point and place a small red dot on it. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The scene shows two circles that move only horizontally toward each other (left-to-right and right-to-left) until they touch and stop. First locate the single tangent point where the two circle boundaries meet, then draw a red circle around that tangent point and place a small red dot on it. Show the complete marking process step by step. Only modify what the task requires; keep the background and any other element unchanged.
Tetris Best Fit Placement
Image sequence
SFT-Image source
Image prompt
A 12×12 Tetris board with row/column labels 0–11: colored cells are locked blocks and the red tetromino is the active piece. It drops into its single best legal placement — clearing the most lines, or leaving the fewest holes if none clear — and completed rows are removed. The piece keeps its red color after it locks. Output one image showing the piece locked in its best placement; if that placement completes any rows, output one more image showing the board after those rows are cleared. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
A 12×12 Tetris board with row/column labels 0–11: colored cells are locked blocks and the red tetromino at the top is the active piece. Animate the red piece rotating and dropping straight down into its single best legal placement — the one clearing the most lines, or (if no move clears a line) leaving the fewest holes — then clear any completed rows. The piece keeps its red color after it locks. Only modify what the task requires; keep the background and any other element unchanged.
Sliding Puzzle
Image sequence
SFT-Image source
Image prompt
Complete this sliding puzzle. The goal is to arrange the numbered tiles in sequential order (filling each row from left to right, with rows from top to bottom), with the blank space at the bottom-right corner. Rules: Only tiles adjacent to the blank space can be moved. Slide one tile per move into the blank space. Complete in exactly 7 moves. Do not make extra moves. Keep the camera view fixed and maintain the grid structure unchanged. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each move, showing the board state after the tile has been slid into the blank space.
Video
Data video source
Video prompt
Complete this sliding puzzle. The goal is to arrange the numbered tiles in sequential order (filling each row from left to right, with rows from top to bottom), with the blank space at the bottom-right corner.
Rules: Only tiles adjacent to the blank space can be moved. Slide one tile per move into the blank space.
Complete in exactly 7 moves.
Do not make extra moves. Keep the camera view fixed and maintain the grid structure unchanged. Only modify what the task requires; keep the background and any other element unchanged.
Color Hue Relative Ordering
Image sequence
SFT-Image source
Image prompt
A 5x5 grid with black borders and labelled rows and columns holds several solid tiles of similar shade arranged in a single connected chain whose hue increases smoothly from one end to the other, joined by a line drawn through them in hue order with a round dot on the start tile and an arrow on the end tile. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the tiles settled into the connected hue-ordered chain with the line drawn through them.
Video
Data video source
Video prompt
Animate the sorting of coloured tiles on a 5x5 lattice. Beginning from a scrambled layout, the tiles step one by one through the empty checkerboard cells and settle into a connected chain along which the hue changes steadily in one direction. A connecting line is then drawn across the chain in hue order, starting from a dot on the first tile and ending in an arrow on the last. The closing frame holds the tidy, hue-ordered chain with its line. Only modify what the task requires; keep the background and any other element unchanged.
Seperate Object Spinning
Image sequence
SFT-Image source
Image prompt
The scene shows 2 objects on the left side and dashed target outlines on the right side. The dashed target outlines remain completely stationary. For each object, first rotate it in place to match the orientation of its corresponding dashed target outline, then move it horizontally to the right so that it aligns exactly with and fits within its corresponding dashed target outline. All objects rotate and translate together at the same time and reach their target outlines simultaneously. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images: the first image shows every object rotated in place to its target orientation (still at its starting position), then the next 3 images show the horizontal translation at uniform progress (1/3, 2/3, ..., 3/3) until each object overlaps its dashed target.
Video
Data video source
Video prompt
The scene shows 2 objects on the left side and dashed target outlines on the right side. The dashed target outlines remain completely stationary. For each object, first rotate it in place to match the orientation of its corresponding dashed target outline, then move it horizontally to the right so that it aligns exactly with and fits within its corresponding dashed target outline. All objects rotate and translate together at the same time and reach their target outlines simultaneously. Only modify what the task requires; keep the background and any other element unchanged.
Visual Jenga
Image sequence
SFT-Image source
Image prompt
The scene shows objects stacked vertically. Extract the objects one by one from top to bottom in order, moving each object out of the frame before extracting the next one. Continue until all objects have been removed from the frame. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each object is removed, showing the remaining stack.
Video
Data video source
Video prompt
The scene shows objects stacked vertically. Extract the objects one by one from top to bottom in order, moving each object out of the frame before extracting the next one. Continue until all objects have been removed from the frame. Only modify what the task requires; keep the background and any other element unchanged.
Attention Shift Same
Image sequence
SFT-Image source
Image prompt
The scene shows two identical square objects, one on the left and one on the right, with a green attention box around the left object. The square objects remain stationary and unchanged throughout. The green attention box moves from the left object to the right object. Only modify what the task requires; keep the background and any other element unchanged. Output a single image showing the final result.
Video
Data video source
Video prompt
The scene shows two identical square objects, one on the left and one on the right, with a green attention box around the left object. The square objects remain stationary and unchanged throughout. Move the green attention box from the left object to the right object. Only modify what the task requires; keep the background and any other element unchanged.
Sliding Number Puzzle Sort
Image sequence
SFT-Image source
Image prompt
A 3x3 sliding number puzzle with tiles 1-8 and one empty cell is shown. Slide the numbered tiles into the empty space so that the numbers are in ascending order from 1 to 8, with the empty cell in the bottom-right corner. Output one image after each sliding move, showing the full sequence of valid moves from the scrambled start to the solved configuration. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
A 3x3 sliding number puzzle with tiles 1-8 and one empty cell is shown. Slide the numbered tiles into the empty space so that the numbers are in ascending order from 1 to 8, with the empty cell in the bottom-right corner. The animation must demonstrate the full sequence of valid sliding moves from the scrambled start configuration to the solved configuration.
Planar Warp Verification
Image sequence
SFT-Image source
Image prompt
Transform the blue grid by aligning its four corners to the four corners of the red quadrilateral. Apply a perspective transformation so the grid matches the red outline. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing the warp transformation at uniform progress intervals.
Video
Data video source
Video prompt
Transform the blue grid by aligning its four corners to the four corners of the red quadrilateral. Apply a perspective transformation so the grid matches the red outline. Only modify what the task requires; keep the background and any other element unchanged. Output the transformed grid.
Shape To Hole Matching
Image sequence
SFT-Image source
Image prompt
Match the 3D objects to the holes and put them in. Only modify what the task requires; keep the background and any other element unchanged. Output 3 images, one after each object is placed: image i shows the first i objects seated in their matching holes.
Video
Data video source
Video prompt
Place each object into the hole with the matching shape. Only modify what the task requires; keep the background and any other element unchanged.
Checker Move
Image sequence
SFT-Image source
Image prompt
You are given a horizontal track of cells with orange balls, red balls, and exactly one empty cell. orange balls travel only right and red balls only left; each move is either a slide into the adjacent empty cell or a jump over a single opposite-coloured ball into the empty cell just past it. The two colours must end with mirrored sides — red on the left, orange on the right, empty in the middle. Output 3 images showing the board after each move, in order. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
You are given a horizontal track of cells with orange balls, red balls, and exactly one empty cell. orange balls travel only right and red balls only left; each move is either a slide into the adjacent empty cell or a jump over a single opposite-coloured ball into the empty cell just past it. The two colours must end with mirrored sides — red on the left, orange on the right, empty in the middle. Generate a video that reveals the moves one by one from the current position. Only modify what the task requires; keep the background and any other element unchanged.
Ball Eating
Image sequence
SFT-Image source
Image prompt
In the scene, there is a black ball and several colored balls of different sizes. The black ball can eat balls that are smaller than itself. After eating a ball, the black ball grows larger. Find the correct sequence to eat all colored balls step by step. Only modify what the task requires; keep the background and any other element unchanged. Output one image after each eating event, showing the updated state.
Video
Data video source
Video prompt
In the scene, there is a black ball and several colored balls of different sizes. The black ball can eat balls that are smaller than itself. After eating a ball, the black ball grows larger. Find the correct sequence to eat all colored balls step by step. Only modify what the task requires; keep the background and any other element unchanged.
Separate Objects No Spin
Image sequence
Data image source
Image prompt
The scene shows 3 objects on the left side and dashed target outlines on the right side. The dashed target outlines remain completely stationary. Move each object horizontally to the right so that it aligns exactly with and fits within its corresponding dashed target outline. All objects move together at the same time and reach their target outlines simultaneously. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images showing all objects at uniform movement progress.
Video
Data video source
Video prompt
The scene shows 3 objects on the left side and dashed target outlines on the right side. The dashed target outlines remain completely stationary. Move each object horizontally to the right so that it aligns exactly with and fits within its corresponding dashed target outline. All objects move together at the same time and reach their target outlines simultaneously. Only modify what the task requires; keep the background and any other element unchanged.
Attention Shift Different
Image sequence
SFT-Image source
Image prompt
The scene shows two objects, one on the left and one on the right, with a green attention box around the left object. The objects remain stationary and unchanged throughout. Move the green attention box from the left object to the right object. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
The scene shows two objects, one on the left and one on the right, with a green attention box around the left object. The objects remain stationary and unchanged throughout. Move the green attention box from the left object to the right object. Only modify what the task requires; keep the background and any other element unchanged.
Shape Sorter
Image sequence
Data image source
Image prompt
Solve the shape sorter puzzle exactly as shown. Move each colored shape card from the left staging area into its matching outline on the right. Keep the camera fixed in a top-down view throughout. Move only one card at a time, sliding each card smoothly without teleportation. Match the pink hexagon, red diamond, and finally the yellow star card. Continue until every outline is filled with its matching colored shape. Only modify what the task requires; keep the background and any other element unchanged. Output two images per card: one showing the card mid-move, and one after it is placed.
Video
Data video source
Video prompt
Solve the shape sorter puzzle exactly as shown. Move each colored shape card from the left staging area into its matching outline on the right. Keep the camera fixed in a top-down view throughout. Move only one card at a time, sliding each card smoothly without teleportation. Match the pink hexagon, red diamond, and finally the yellow star card. Continue until every outline is filled with its matching colored shape. Only modify what the task requires; keep the background and any other element unchanged.
Wheel Of Fortune
Image sequence
SFT-Image source
Image prompt
You are given a wheel of fortune: a circle of equal-sized colored sectors, each holding a different prize, with a triangular marker fixed at the top. From the position shown, the wheel turns CLOCKWISE by 2.50 revolutions before stopping. Report the prize that finishes under the marker. Output 1 image showing the final state with the solution. Only modify what the task requires; keep the background and any other element unchanged.
Video
Data video source
Video prompt
You are given a wheel of fortune: a circle of equal-sized colored sectors, each holding a different prize, with a triangular marker fixed at the top. From the position shown, the wheel turns CLOCKWISE by 2.50 revolutions before stopping. Report the prize that finishes under the marker. Generate a video that shows the wheel spinning clockwise, slowing to a stop on the winning sector. Only modify what the task requires; keep the background and any other element unchanged.
Construct Concentric Ring
Image sequence
SFT-Image source
Image prompt
The scene shows two circles of different sizes at different positions.
Keep both circles unchanged in size and color. Only move their positions.
Move both circles until they share the same center point (concentric circles) exactly at the center of the image. Only modify what the task requires; keep the background and any other element unchanged.
Output one image showing the final result.
Video
Data video source
Video prompt
The scene shows two circles of different sizes at different positions.
Keep both circles unchanged in size and color. Only move their positions.
Move both circles until they share the same center point (concentric circles) exactly at the center of the image. Only modify what the task requires; keep the background and any other element unchanged.
Return To Correct Bin
Image sequence
SFT-Image source
Image prompt
Move each item into the bin that matches its color. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result.
Video
Data video source
Video prompt
Move each item into the bin that matches its color. Only modify what the task requires; keep the background and any other element unchanged.
2D Object Rotation
Image sequence
Data image source
Image prompt
The scene contains 1 2D object(s). Show them rotating clockwise by 35 degrees around their respective centroids. Only modify what the task requires; keep the background and any other element unchanged. Output 4 images at uniform rotation progress (1/4, 2/4, ..., 4/4 of the total rotation).
Video
Data video source
Video prompt
The scene contains 1 2D object(s). Show them rotating clockwise by 35 degrees around their respective centroids. Only modify what the task requires; keep the background and any other element unchanged.
Symbol Substitute
Image sequence
Data image source
Image prompt
Substitute the solid heart at position 2 with a orange solid left triangle. A reference panel in the top-right shows the target symbol. Only modify what the task requires; keep the background and any other element unchanged. Output 1 image showing the final result, in which the old symbol has been removed and the new symbol appears at the same position in its full color, with every other position left unchanged.
Video
Data video source
Video prompt
Substitute the solid heart at position 2 with a orange solid left triangle. A reference panel in the top-right shows the target symbol. The animation shows the old symbol smoothly fading out into the white background, then the new symbol smoothly fading in from white to its full color at the same position. Only modify what the task requires; keep the background and any other element unchanged.
Circle All Squares From Mixed Shapes
Image sequence
SFT-Image source
Image prompt
The scene shows 3 axis-aligned squares and rectangles. A square has equal width and height, while a rectangle does not. First identify all squares, then draw one red circle around each square. Do not circle any rectangles. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The scene shows 3 axis-aligned squares and rectangles. A square has equal width and height, while a rectangle does not. First identify all squares, then draw one red circle around each square. Do not circle any rectangles. Show the complete circling process step by step. Only modify what the task requires; keep the background and any other element unchanged.
Identify All Hollow Points
Image sequence
SFT-Image source
Image prompt
The image contains multiple points, including solid filled circles and hollow outlined circles. First identify every hollow point (outlined and not filled), then circle each hollow point with a red ring. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The image contains multiple points, including solid filled circles and hollow outlined circles. First identify every hollow point (outlined and not filled), then circle each hollow point with a red ring. Show the complete solution step by step. Only modify what the task requires; keep the background and any other element unchanged.
Arrange Circles By Circumference
Image sequence
Data image source
Image prompt
The scene shows 7 circles of different sizes and colors arranged randomly.
Keep every circle unchanged in size and color. Only rearrange their positions.
Align all circles on a single horizontal line. Center the entire row of circles in the image.
Sort them from left to right by circumference, from largest to smallest. Only modify what the task requires; keep the background and any other element unchanged. Output one image showing the final result.
Video
Data video source
Video prompt
The scene shows 7 circles of different sizes and colors arranged randomly.
Keep every circle unchanged in size and color. Only rearrange their positions.
Align all circles on a single horizontal line. Center the entire row of circles in the image.
Sort them from left to right by circumference, from largest to smallest. Only modify what the task requires; keep the background and any other element unchanged.