Datasets:

Video-Reason
/

VBVR-Pro-Bench

task stringlengths 38 89	sample_id stringclasses 2 values	prompt stringlengths 128 932
In-Domain_50/G-131_select_next_figure_increasing_size_sequence_data-generator	00000	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 (o...
In-Domain_50/G-131_select_next_figure_increasing_size_sequence_data-generator	00001	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 (o...
In-Domain_50/G-134_select_next_figure_large_small_alternating_sequence_data-generator	00000	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 b...
In-Domain_50/G-134_select_next_figure_large_small_alternating_sequence_data-generator	00001	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 b...
In-Domain_50/G-138_spot_unique_non_repeated_color_data-generator	00000	The scene shows 5 geometric shapes with different colors. Exactly one color appears only once, while every other color appears at least twice. First find the shape whose color is unique, then outline that entire shape with a black contour. Do not add any extra marks. Show the complete marking process step by step. Only...
In-Domain_50/G-138_spot_unique_non_repeated_color_data-generator	00001	The scene shows 3 geometric shapes with different colors. Exactly one color appears only once, while every other color appears at least twice. First find the shape whose color is unique, then outline that entire shape with a black contour. Do not add any extra marks. Show the complete marking process step by step. Only...
In-Domain_50/G-13_grid_number_sequence_data-generator	00000	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.
In-Domain_50/G-13_grid_number_sequence_data-generator	00001	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.
In-Domain_50/G-158_identify_all_hollow_points_data-generator	00000	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 eleme...
In-Domain_50/G-158_identify_all_hollow_points_data-generator	00001	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 eleme...
In-Domain_50/G-15_grid_avoid_obstacles_data-generator	00000	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 t...
In-Domain_50/G-15_grid_avoid_obstacles_data-generator	00001	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 t...
In-Domain_50/G-16_grid_go_through_block_data-generator	00000	The scene shows a 10×10 grid with a green start cell, a red end cell, and colored block cells (brown, blue, purple). 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 (brown, blue, purple) in order, taking the shortest path ...
In-Domain_50/G-16_grid_go_through_block_data-generator	00001	The scene shows a 10×10 grid with a green start cell, a red end cell, and colored block cells (blue, purple, pink, 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, purple, pink, yellow) in order, taking the ...
In-Domain_50/G-18_grid_shortest_path_data-generator	00000	The scene shows a 10x10 grid with a orange start square (containing a green circular agent) and a red end square. Starting from the orange 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. Only modify what the task re...
In-Domain_50/G-18_grid_shortest_path_data-generator	00001	The scene shows a 10x10 grid with a green start square (containing a yellow circular agent) and a pink end square. Starting from the green 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 r...
In-Domain_50/G-194_construct_concentric_ring_data-generator	00000	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 a...
In-Domain_50/G-194_construct_concentric_ring_data-generator	00001	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 a...
In-Domain_50/G-21_multiple_occlusions_vertical_data-generator	00000	The scene shows 5 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....
In-Domain_50/G-21_multiple_occlusions_vertical_data-generator	00001	The scene shows 6 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....
In-Domain_50/G-25_seperate_object_spinning_data-generator	00000	The scene shows 4 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 exactl...
In-Domain_50/G-25_seperate_object_spinning_data-generator	00001	The scene shows 3 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 exactl...
In-Domain_50/G-29_chart_extreme_with_data_data-generator	00000	The scene shows a pie chart with numeric data labels. Find the maximum value segment and draw a red outline border around the corresponding segment to highlight it. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/G-29_chart_extreme_with_data_data-generator	00001	The scene shows a scatter 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.
In-Domain_50/G-31_directed_graph_navigation_data-generator	00000	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 no...
In-Domain_50/G-31_directed_graph_navigation_data-generator	00001	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 no...
In-Domain_50/G-39_attention_shift_different_data-generator	00000	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 eleme...
In-Domain_50/G-39_attention_shift_different_data-generator	00001	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 eleme...
In-Domain_50/G-3_stable_sort_data-generator	00000	The scene contains two types of shapes, each type has three shapes of different sizes arranged randomly. Keep all shapes unchanged in appearance (type, size, and color). Only rearrange their positions: first group the shapes by type, then within each group, sort the shapes from smallest to largest (left to right), and ...
In-Domain_50/G-3_stable_sort_data-generator	00001	The scene contains two types of shapes, each type has three shapes of different sizes arranged randomly. Keep all shapes unchanged in appearance (type, size, and color). Only rearrange their positions: first group the shapes by type, then within each group, sort the shapes from smallest to largest (left to right), and ...
In-Domain_50/G-41_grid_highest_cost_data-generator	00000	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 cel...
In-Domain_50/G-41_grid_highest_cost_data-generator	00001	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 cel...
In-Domain_50/G-43_understand_scene_structure_data-generator	00000	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.
In-Domain_50/G-43_understand_scene_structure_data-generator	00001	The scene shows a floorplan with multiple rooms. Identify the living room 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.
In-Domain_50/G-45_key_door_matching_data-generator	00000	[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 ...
In-Domain_50/G-45_key_door_matching_data-generator	00001	[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 ...
In-Domain_50/G-51_predict_next_color_data-generator	00000	Predict the next color in the sequence. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/G-51_predict_next_color_data-generator	00001	Predict the next color in the sequence. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/G-5_multi_object_placement_data-generator	00000	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 t...
In-Domain_50/G-5_multi_object_placement_data-generator	00001	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 t...
In-Domain_50/G-8_track_object_movement_data-generator	00000	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.
In-Domain_50/G-8_track_object_movement_data-generator	00001	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.
In-Domain_50/G-9_identify_objects_in_region_data-generator	00000	Outline all circles in the circular region with a green border. Only outline objects within that region. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/G-9_identify_objects_in_region_data-generator	00001	Outline all trapezoids in the square region with a green border. Only outline objects within that region. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/O-10_shape_outline_fill_data-generator	00000	Follow the visual pattern A:B :: C:? where the first row shows a style change between two identical shapes. Animate the second-row right shape so that its fill or outline style matches the same transformation. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/O-10_shape_outline_fill_data-generator	00001	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.
In-Domain_50/O-12_shape_color_then_scale_data-generator	00000	The scene shows an analogy A→B→C :: D→?→? with two rows of shapes. Top row: color_155 circle at medium becomes color_246 circle at medium, then color_246 circle at small. This shows two rules: first change color from color_155 to color_246, then change size from medium to small. Bottom row: oval starts with color_155 a...
In-Domain_50/O-12_shape_color_then_scale_data-generator	00001	The scene shows an analogy A→B→C :: D→?→? with two rows of shapes. Top row: color_316 heart at large becomes color_118 heart at large, then color_118 heart at small. This shows two rules: first change color from color_316 to color_118, then change size from large to small. Bottom row: minus starts with color_316 and la...
In-Domain_50/O-13_shape_outline_then_move_data-generator	00000	The scene shows an analogy A→B→C :: D→?→? with two rows of shapes and arrows. On the top row, a filled arrow first becomes an outline-only arrow (step 1), then moves down (step 2). On the bottom row, the star starts filled. Apply the same two-step transformation: first convert it to outline-only style, then move it dow...
In-Domain_50/O-13_shape_outline_then_move_data-generator	00001	The scene shows an analogy A→B→C :: D→?→? with two rows of shapes and arrows. On the top row, a filled rhombus first becomes an outline-only rhombus (step 1), then moves down (step 2). On the bottom row, the crescent starts filled. Apply the same two-step transformation: first convert it to outline-only style, then mov...
In-Domain_50/O-14_shape_scale_then_outline_data-generator	00000	Show the shape first changing size and then becoming outline-only. Both transformations should match the example pattern. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/O-14_shape_scale_then_outline_data-generator	00001	Show the shape first changing size and then becoming outline-only. Both transformations should match the example pattern. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/O-15_ball_bounces_given_time_data-generator	00000	A ball is placed at the initial position with a direction arrow indicating its movement direction. Simulate the ball bouncing 2 times off the boundary walls following elastic collision physics (angle of incidence equals angle of reflection). The ball stops after the 2th bounce, with its final position at the wall where...
In-Domain_50/O-15_ball_bounces_given_time_data-generator	00001	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...
In-Domain_50/O-16_color_addition_data-generator	00000	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 modi...
In-Domain_50/O-16_color_addition_data-generator	00001	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 modi...
In-Domain_50/O-18_glass_refraction_data-generator	00000	A horizontal air–glass interface is drawn with the glass region hatched below. A blue incident ray reaches the interface from air at θ = 69.8° from the normal (n_air = 1.00, n_glass = 1.850). Draw the red refracted ray inside the glass according to Snell's law, extending it to the image boundary. The angle annotation s...
In-Domain_50/O-18_glass_refraction_data-generator	00001	A horizontal air–glass interface is drawn with the glass region hatched below. A blue incident ray reaches the interface from air at θ = 41.7° from the normal (n_air = 1.00, n_glass = 1.387). Draw the red refracted ray inside the glass according to Snell's law, extending it to the image boundary. The angle annotation s...
In-Domain_50/O-19_mirror_reflection_data-generator	00000	Given the mirror reflectivity = 0.71, 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.
In-Domain_50/O-19_mirror_reflection_data-generator	00001	Given the mirror reflectivity = 0.46, predict the reflection of light when it hits the mirror. The reflected ray should extend to the edge of the image. The angle annotation should be removed. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/O-21_construction_blueprint_data-generator	00000	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 match...
In-Domain_50/O-21_construction_blueprint_data-generator	00001	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 match...
In-Domain_50/O-23_domino_chain_branch_path_prediction_data-generator	00000	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 light gray trunk falls first, then splits into the pink Branch A (upward) and cyan Branch B (downward). Both branches complete successfully. The final state shows all ...
In-Domain_50/O-23_domino_chain_branch_path_prediction_data-generator	00001	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 light gray trunk falls first, then splits into the red Branch A (upward) and blue Branch B (downward). Branch B has a gap marked with a red X that stops the chain reac...
In-Domain_50/O-24_domino_chain_gap_analysis_data-generator	00000	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 un...
In-Domain_50/O-24_domino_chain_gap_analysis_data-generator	00001	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 un...
In-Domain_50/O-25_LEGO_construction_assembly_data-generator	00000	This is LEGO assembly step 3. The scene shows a partial model on the right and a 2x2 red brick on the left in a callout box. Take the red 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. ...
In-Domain_50/O-25_LEGO_construction_assembly_data-generator	00001	This is LEGO assembly step 4. The scene shows a partial model on the right and a 2x4 blue brick on the left in a callout box. Take the blue 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...
In-Domain_50/O-29_ballcolor_data-generator	00000	In the scene, there are 4 ball clusters: red cluster A has 3 balls, and there are other clusters with different numbers of balls. The global TOTAL label shows the sum 14. 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 ...
In-Domain_50/O-29_ballcolor_data-generator	00001	In the scene, there are 4 ball clusters: red cluster A has 3 balls, and there are other clusters with different numbers of balls. The global TOTAL label shows the sum 14. 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 ...
In-Domain_50/O-30_bookshelf_data-generator	00000	In the scene, there is a bookshelf with a set of books already placed, and a few pink books waiting on the right. There are gaps between the books on the shelf. Place each pink 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 requir...
In-Domain_50/O-30_bookshelf_data-generator	00001	In the scene, there is a bookshelf with a set of books already placed, and a few yellow books waiting on the right. There are gaps between the books on the shelf. Place each yellow 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 re...
In-Domain_50/O-31_ball_eating_data-generator	00000	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 ...
In-Domain_50/O-31_ball_eating_data-generator	00001	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 ...
In-Domain_50/O-32_rolling_ball_data-generator	00000	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...
In-Domain_50/O-32_rolling_ball_data-generator	00001	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...
In-Domain_50/O-33_counting_object_data-generator	00000	The scene displays various geometric objects scattered across the image. Each object is a filled shape with a black outline. Starting from any position in the image, systematically count all visible objects in the scene, regardless of their shape or color. Count each object exactly once, highlighting its border when co...
In-Domain_50/O-33_counting_object_data-generator	00001	The scene displays various geometric objects scattered across the image. Each object is a filled shape with a black outline. Starting from any position in the image, systematically count all visible objects in the scene, regardless of their shape or color. Count each object exactly once, highlighting its border when co...
In-Domain_50/O-34_dot_to_dot_task_data-generator	00000	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.
In-Domain_50/O-34_dot_to_dot_task_data-generator	00001	The scene shows 5 numbered dots scattered across the image. Connect the dots in numerical order (1→2→3→...→5) 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.
In-Domain_50/O-36_grid_shift_data-generator	00000	The scene shows a 9x9 grid with 27 purple square blocks, each with a black outline, positioned at various locations. All blocks must move simultaneously rightward by exactly 2 steps. Each block shifts one grid cell per step toward the right, and all blocks must remain within the grid boundaries throughout the movement....
In-Domain_50/O-36_grid_shift_data-generator	00001	The scene shows a 8x8 grid with 4 teal square blocks, each with a black outline, positioned at various locations. All blocks must move simultaneously leftward by exactly 2 steps. Each block shifts one grid cell per step toward the left, and all blocks must remain within the grid boundaries throughout the movement. Afte...
In-Domain_50/O-37_light_sequence_data-generator	00000	The scene shows 10 circular lights in a horizontal row on a white background. Lights on are gold/yellow with glow; lights off are gray. Initially, some lights are on and some are off. Your task: Modify the light states so that the 6th light from the left (counting from left to right) is on (gold/yellow with glow), and...
In-Domain_50/O-37_light_sequence_data-generator	00001	The scene shows 8 circular lights in a horizontal row on a white background. Lights on are gold/yellow with glow; lights off are gray. Initially, some lights are on and some are off. Your task: Modify the light states so that all lights at odd positions (counting from left to right: 1st is odd, 2nd is even, 3rd is odd...
In-Domain_50/O-38_majority_color_data-generator	00000	The image shows multiple colored objects. Count how many objects are of each color. Identify the majority color (the color with the most objects). Then make all objects of other colors vanish, while the majority color objects remain stationary. Only modify what the task requires; keep the background and any other eleme...
In-Domain_50/O-38_majority_color_data-generator	00001	The image shows multiple colored objects. Count how many objects are of each color. Identify the majority color (the color with the most objects). Then make all objects of other colors vanish, while the majority color objects remain stationary. Only modify what the task requires; keep the background and any other eleme...
In-Domain_50/O-44_rotation_puzzle_data-generator	00000	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 reac...
In-Domain_50/O-44_rotation_puzzle_data-generator	00001	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 reac...
In-Domain_50/O-45_sequence_completion_data-generator	00000	The scene shows a color_cycle sequence. Elements are arranged horizontally from left to right. The last position contains a question mark (?) indicating a missing element. Observe the pattern: the colors follow a cyclic order that repeats after a certain number of elements. Determine the element that should replace the...
In-Domain_50/O-45_sequence_completion_data-generator	00001	The scene shows a color_cycle sequence. Elements are arranged horizontally from left to right. The last position contains a question mark (?) indicating a missing element. Observe the pattern: the colors follow a cyclic order that repeats after a certain number of elements. Determine the element that should replace the...
In-Domain_50/O-47_sliding_puzzle_data-generator	00000	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. Co...
In-Domain_50/O-47_sliding_puzzle_data-generator	00001	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. Co...
In-Domain_50/O-52_traffic_light_data-generator	00000	This scene shows a crossroad with four traffic lights (North, South, East, West). Each light independently follows a 3-color cycle: Red (4s) -> Yellow (4s) -> Green (4s) -> Yellow (4s) -> Red. Currently: North light is yellow with 1s countdown, South light is yellow with 2s countdown, East light is yellow with 4s count...
In-Domain_50/O-52_traffic_light_data-generator	00001	This scene shows a crossroad with four traffic lights (North, South, East, West). Each light independently follows a 3-color cycle: Red (4s) -> Yellow (4s) -> Green (4s) -> Yellow (4s) -> Red. Currently: North light is red with 4s countdown, South light is green with 4s countdown, East light is green with 3s countdown,...
In-Domain_50/O-53_clock_data-generator	00000	The clock shows 7:44. Show what the clock will look like after 17 hours. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/O-53_clock_data-generator	00001	The clock shows 10:04. Show what the clock will look like after 11 hours. Only modify what the task requires; keep the background and any other element unchanged.
In-Domain_50/O-55_rotation_data-generator	00000	A 6-block sculpture sits fixed on a table. First frame: Your camera is tilted at 31° elevation, viewing from 140° azimuth. Final frame: Your camera remains at 31° elevation, but rotates horizontally to 320° azimuth. This is a 180-degree rotation Create a smooth video showing the camera's horizontal rotation around the ...
In-Domain_50/O-55_rotation_data-generator	00001	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 s...
In-Domain_50/O-75_communicating_vessels_data-generator	00000	A system of 3 communicating vessels with equal-diameter vertical tubes is filled with oil (oil-like (medium viscosity)), which appears yellow in color. As shown in the initial frame, the liquid levels in the tubes are [37, 54, 59] cm respectively. Due to pressure differences between the tubes, the liquid begins to flow...
In-Domain_50/O-75_communicating_vessels_data-generator	00001	A system of 5 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 [29, 6, 19, 47, 37] cm respectively. Due to pressure differences between the tubes, the liquid begin...

End of preview. Expand in Data Studio

VBVR-Pro-Bench

The frozen evaluation split of VBVR-Pro. 100 parameterized reasoning tasks, 5 held-out instances each, in both a video (I2V) and an interleaved-image setting.

At a glance

Property	Value
Tasks	100 (`In-Domain_50` 50 + `Out-of-Domain_50` 50)
Instances per task	5
Total instances	500 per setting (1,000 across both)
Settings	Video and Image
Resolution	1024 × 1024
Video	16 fps, MPEG-4 Part 2, as generated (no re-encode)
Size	86.0 MB (video) + 11.7 MB (image)

In-Domain_50 covers task families that also appear in the training splits; Out-of-Domain_50 covers task families that do not, and is the split to report generalization on.

Layout

.
├── VBVR-Pro-Bench-Video.tar.gz
└── VBVR-Pro-Bench-Image.tar.gz

VBVR-Pro-Bench-Video.tar.gz extracts to:

VBVR-Pro-Bench-Video/
├── In-Domain_50/
│   ├── G-131_select_next_figure_increasing_size_sequence_data-generator/
│   │   ├── 00000/
│   │   │   ├── first_frame.png        # conditioning image  (1024 × 1024)
│   │   │   ├── prompt.txt             # instruction
│   │   │   ├── ground_truth.mp4       # reference video     (16 fps)
│   │   │   ├── final_frame.png        # last frame of the reference video
│   │   │   └── metadata.json          # task parameters, ground truth, scoring contract
│   │   ├── 00001/ … 00004/            # same five files
│   │   └── …
│   └── …                              # 50 tasks
└── Out-of-Domain_50/                  # 50 tasks, same shape

VBVR-Pro-Bench-Image.tar.gz extracts to the same tree, with the reference output as an image sequence instead:

VBVR-Pro-Bench-Image/
├── In-Domain_50/
│   ├── G-131_select_next_figure_increasing_size_sequence_data-generator/
│   │   ├── 00000/
│   │   │   ├── first_frame.png        # conditioning image  (1024 × 1024)
│   │   │   ├── prompt.txt             # instruction
│   │   │   ├── frame_1.png            # reference output, step 1
│   │   │   ├── frame_2.png            # … step 2, when the task has more steps
│   │   │   ├── …                      # up to frame_N.png
│   │   │   └── metadata.json          # task parameters, ground truth, scoring contract
│   │   ├── 00001/ … 00004/
│   │   └── …
│   └── …                              # 50 tasks
└── Out-of-Domain_50/                  # 50 tasks, same shape

N is task-dependent (1–21): it is the number of steps the instruction asks the model to render, and it is stated in prompt.txt (e.g. "Output 4 images: …"). Most tasks are single-step.

Usage

huggingface-cli download Video-Reason/VBVR-Pro-Bench --repo-type dataset --local-dir ./VBVR-Pro-Bench
cd VBVR-Pro-Bench && tar xzf VBVR-Pro-Bench-Video.tar.gz

Generate one video (or image sequence) per instance from first_frame.png + prompt.txt, then score with VBVR-Pro-Bench.

License

VBVR-Pro source code, scripts, configuration files and task-specific scoring software — including everything in this repository — are licensed under the Apache License 2.0. VBVR-Pro data and benchmark materials are separately licensed under CC BY-NC 4.0. Model weights and third-party materials remain subject to their applicable model-card and upstream terms. See LICENSE.md for details.

Citation

@misc{xu2026vbvrproscalableverifiablesuite,
      title={VBVR-Pro: A Scalable and Verifiable Suite for Native Visual Reasoning},
      author={Junxiang Xu and Ruisi Wang and Fanyi Pu and Maijunxian Wang and Ran Ji and Tongxi Zhou and Chenyang Gu and Jing Zuo and Hongcan Xiao and Yimeng Geng and Wanqi Yin and Wei Chen and Oscar Qian and Zhengan Yan and Ziqi Huang and Haiwen Diao and Liang Pan and Bo Li and Xiangyu Fan and Dezhi Luo and Fengyuan Yu and Zehong Zhao and Qingying Gao and Tinghui Zhu and Yilan Zhang and Jingqi Tong and Pinyuan Feng and Zhengze Jiang and Letian Wang and Ziyu Guo and Renrui Zhang and Jieneng Chen and Sonia Joseph and Constantin Venhoff and Saman Motamed and Mengyue Yang and Chandra Sripada and Alan Yuille and Philip Torr and Lvmin Zhang and Vikash Kumar and Daniel Khashabi and Nikolaus Kriegeskorte and Rapha\"{e}l Milli\`{e}re and Vincent C. M\"{u}ller and Anyi Rao and Quan Wang and Ziwei Liu and Dahua Lin and Lei Yang and Hokin Deng and Zhongang Cai},
      year={2026},
      eprint={2608.26105},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2608.26105},
}

Downloads last month: 7

Collection including Video-Reason/VBVR-Pro-Bench

VBVR-Pro: A Scalable and Verifiable Suite for Native Visual

Collection

22 items • Updated about 8 hours ago • 4

Paper for Video-Reason/VBVR-Pro-Bench

VBVR-Pro: A Scalable and Verifiable Suite for Native Visual Reasoning

Paper • 2608.26105 • Published 1 day ago • 32