Pixqueeze is a web-based image scaling application designed for high-quality downscaling of low-resolution images and pixel art. It uses advanced algorithms to preserve features and clarity that standard resizing methods often lose.
Pixqueeze offers a variety of scaling methods, from standard techniques to our custom-built algorithms:
- Nearest Neighbor: The classic approach. Fast and preserves the raw, blocky look of pixels without any blurring.
- Bicubic Interpolation: Standard smooth scaling, useful for natural photographic images where soft transitions are desired.
- Palette-Aware Area: A smart scaling method that first analyzes the image's palette. It "votes" on the final color of a pixel based on the most prominent palette colors in the source area. This is excellent for maintaining distinct colors in pixel art without the muddiness of standard interpolation.
- Edge-Priority Scaler (formerly Contour): A content-aware scaling method. Instead of blindly averaging pixels, it scans the source area for high-contrast "seeds"—pixels that define edges or details. It then grows a region around these seeds to determine the final pixel color. This ensures that fine lines and significant details are preserved and not washed out by the background.
- Sharpener: Our flagship high-fidelity pipeline. It combines the Edge-Priority logic with a multi-stage post-processing pass:
- Smart Scaling: Uses edge-priority logic to maintain structure.
- Bilateral Filtering: Smooths out noise while strictly preserving edge sharpness.
- Palette Optimization: Dynamically reduces the color count by merging similar shades.
- Snap: Forces pixels to align with the optimized palette, producing a clean, posterized look perfect for cleaning up noisy scans or sketches.
Enhance your results further with integrated processing steps:
- Bilateral Filter: An edge-preserving smoothing filter that reduces noise without blurring important boundary details.
- Wavelet Sharpening: Intelligent sharpening that boosts high-frequency details without overshooting.
- Palette Optimization: Automatically optimizes the image's colors by merging similar shades, helping to clean up noisy inputs.
- Web Worker Architecture: All heavy image processing and scaling computations are offloaded to background threads. This ensures the User Interface remains responsive and lag-free, even during intensive operations.
- Real-time Processing: Optimized for fast iteration, providing near-instant feedback as you adjust parameters.
- Multi-Zoom Comparisons: View processed results at 1x, 2x, and 3x zoom levels side-by-side in a centralized grid for easy evaluation.
- Adaptive Processing: The application intelligently scales large inputs to optimal processing sizes.
Downscaled to 32 or 64 pixels respectively.
Samples include nearest neighbor, palette area, bicubic, and our custom scalers with various optimizations.
Pixqueeze is built with a modern, performance-first tech stack:
- Core: React 19 + TypeScript
- Build Tool: Vite
- State Management: Nanostores (Atomic, lightweight state)
- Concurrency: Comlink (Web Workers)
- Styling: Vanilla CSS
- Tooling: Biome (Linting & Formatting), Vitest (Testing)
- Node.js
- pnpm
# Clone the repository
git clone https://github.com/yourusername/pixqueeze.git
# Navigate to project
cd pixqueeze
# Install dependencies
pnpm install# Start development server
pnpm run dev# Generate production bundle
pnpm run buildThis project is licensed under the MIT License - see the LICENSE file for details.
Made by Mendrik