O-9: Shape Scaling Data Generator

Generates synthetic tasks showing an analogy A:B :: C:? with shape scaling transformations. The goal is to infer the scaling factor from A to B and apply the same scaling to C.

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.

---

📌 Basic Information

Property	Value
Task ID	O-9
Task	Shape Scaling
Category	Abstraction
Resolution	1024×1024 px
FPS	16 fps
Duration	~3.8 seconds
Output	PNG images + MP4 video

---

🚀 Usage

Installation

# Clone the repository
git clone https://github.com/VBVR-DataFactory/O-9_shape_scaling_data-generator.git
cd O-9_shape_scaling_data-generator

# Install dependencies
pip install -r requirements.txt

Generate Data

# Generate 100 samples
python examples/generate.py --num-samples 100

# Generate with specific seed
python examples/generate.py --num-samples 100 --seed 42

# Generate without videos
python examples/generate.py --num-samples 100 --no-videos

# Custom output directory
python examples/generate.py --num-samples 100 --output data/my_output

Command-Line Options

Argument	Type	Description	Default
--num-samples	int	Number of samples to generate	100
--seed	int	Random seed for reproducibility	Random
--output	str	Output directory	data/questions
--no-videos	flag	Skip video generation	False

---

📖 Task Example

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 plus is enlarged to 115% of its original size (scaling factor 1.15). 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.

Visual

Initial Frame Analogy A:B :: C:? with scaling rule	Animation Shape C scales by same factor as A→B	Final Frame Scaling rule applied to C

---

📖 Task Description

Objective

Infer the scaling factor from A to B in the analogy, then apply the same scaling to shape C to determine the answer.

Task Setup

• Analogy Format: A:B :: C:? with four positions
• Top Row (Rule): Shows shape A and its scaled version B
• Bottom Row (Query): Shows shape C and unknown answer ?
• Scaling Inference: Determine scaling factor from A to B (e.g., 1.10 = 110%) - this generator modifies the scale parameter of shapes while keeping all other properties (shape type, color, position, rotation) unchanged
• Rule Application: Apply same scaling factor to C
• Uniform Scaling: Both width and height scaled by same factor

Key Features

• Analogical reasoning: Understanding A:B :: C:? relationship pattern
• Scale inference: Extracting scaling factor from example transformation
• Rule application: Applying inferred scaling to new shape
• Proportional scaling: Maintaining aspect ratio during transformation
• Factor preservation: Maintaining exact scaling factor across different shapes
• Abstract reasoning: Generalizing scaling transformation across different geometries
• Large-scale diversity: 20 distinct shapes, 36 scale factors, and 400 colors for extensive variety (5.4M+ unique combinations)

---

📦 Data Format

data/questions/shape_scaling_task/shape_scaling_00000000/
├── first_frame.png      # Initial state (analogy with query)
├── final_frame.png      # Goal state (query shape scaled)
├── prompt.txt           # Task instructions
├── ground_truth.mp4     # Solution video (16 fps)
└── question_metadata.json # Task metadata

File specifications: Images are 1024×1024 PNG. Videos are MP4 at 16 fps, approximately 3.8 seconds long.

---

🏷️ Tags

analogical-reasoning scaling-transformation visual-analogy scale-inference geometric-transformation abstract-reasoning proportional-scaling

---