A high-performance image processing tool that simplifies images using adaptive quad-tree partitioning with variance-based splitting.

This is a Rust command-line tool that transforms images into a simplified representation by:

• Dividing the image into rectangular regions of varying sizes
• Using more detail (smaller rectangles) where the image is complex
• Using less detail (larger rectangles) where the image is simple

The program analyzes your image and creates a quad-tree structure:

1. Starts with the whole image as one rectangle
2. Finds the region with the most color variation (faces, edges, text, patterns)
3. Splits that region into 4 quadrants (northwest, northeast, southwest, southeast)
4. Repeats for N iterations
5. Fills each final region with its average color

1000000 iterations

50000 iterations

25000 iterations

10000 iterations

1000 iterations

• Rust (version 1.70 or later)

cd qtree
cargo build --release

cargo run --release -- <input-image> -iter <iterations> [options]

• <input-image> - Path to your input image (supports .jpg, .png, .gif, etc.)
• -iter <N> - Number of subdivisions to perform
• -o <path> - Output file path (default: input-compressed.ext in current directory)
• -outline <hex> - Add colored borders around regions (e.g., 000000 for black, FF0000 for red)
• -gif <interval> - Create animated GIF showing the algorithm's progress, saving every N iterations
• -outline HEXCOLOR - Draw borders around each region, making the quad-tree structure visible.

All dependencies are automatically downloaded by Cargo.