Generates synthetic datasets for training and evaluating vision models on sequential object removal and physics reasoning tasks. Each sample contains vertically stacked objects that must be extracted from top to bottom in order.
Each sample pairs a task (first frame + prompt describing what needs to happen) with its ground truth solution (final frame showing the result + video demonstrating how to achieve it). This structure enables both model evaluation and training.
| Property | Value |
|---|---|
| Task ID | G-33 |
| Task | Visual Jenga |
| Category | Spatiality |
| Resolution | 1024×1024 px |
| FPS | 16 fps |
| Duration | ~5-6 seconds |
| Output | PNG images + MP4 video |
# 1. Clone the repository
git clone https://github.com/VBVR-DataFactory/G-33_visual_jenga_data-generator.git
cd G-33_visual_jenga_data-generator
# 2. Create and activate virtual environment
python3 -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# 3. Install dependencies
pip install --upgrade pip
pip install -r requirements.txt
pip install -e .# Generate 50 samples
python examples/generate.py --num-samples 50
# Custom output directory
python examples/generate.py --num-samples 100 --output data/my_dataset
# Reproducible generation with seed
python examples/generate.py --num-samples 50 --seed 42
# Without videos (faster)
python examples/generate.py --num-samples 50 --no-videos| Argument | Description |
|---|---|
--num-samples |
Number of tasks to generate (required) |
--output |
Output directory (default: data/questions) |
--seed |
Random seed for reproducibility |
--no-videos |
Skip video generation (images only) |
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.
 |
 |
 |
| Initial Frame Objects stacked vertically |
Animation Objects removed top to bottom |
Final Frame All objects removed, empty frame |
Remove vertically stacked objects sequentially from top to bottom, extracting each one completely before proceeding to the next.
- Stack configuration: Multiple objects stacked vertically
- Object types: Various shapes (rectangles, circles, triangles)
- Extraction order: Top to bottom (strict sequential order)
- Removal method: Move each object completely out of frame
- Sequential constraint: Must finish removing one object before starting next
- Background: White with clear visibility
- Goal: Remove all objects in correct order until frame is empty
- Sequential top-to-bottom extraction order
- Physics-based stacking visualization
- Complete removal of each object before next
- Tests understanding of vertical ordering
- Jenga-like sequential unstacking logic
- Smooth animation showing extraction process
data/questions/visual_jenga_task/visual_jenga_00000000/
├── first_frame.png # Initial stack of objects
├── final_frame.png # Empty frame after all removals
├── prompt.txt # Sequential extraction instruction
├── ground_truth.mp4 # Animation of top-to-bottom removal
└── question_metadata.json # Task metadata
File specifications:
- Images: 1024×1024 PNG format
- Video: MP4 format, 16 fps
- Duration: ~5-6 seconds (depends on number of objects)
physics sequential-reasoning object-removal spatial-ordering stacking procedural-reasoning