AI-powered animation sprite generation studio for game developers, addressing the challenge of AI models producing inconsistent and engine-unfriendly sprite sheets. PerfectPixel Studio automates the creation of 100+ animations and 8-direction sprite sets from text prompts, ensuring technical precision and quality required for game engines.

How It Works

This project employs a hybrid approach: AI generates initial character assets, followed by a deterministic post-processing pipeline that enforces quality, consistency, and precision. The core innovation lies in applying signal processing techniques across three axes: robust YCbCr chrominance-based matting for background removal, projection profile and dynamic programming for accurate frame segmentation, and alpha-weighted centroids for stable frame alignment. This pipeline includes a self-correcting quality loop with measurement-driven retry hints, ensuring results converge to a high, consistent quality bar, overcoming AI non-determinism.

Quick Start & Requirements

• Primary install/run: Requires Go 1.25+, Node.js 18+, and Wails CLI v2 (go install github.com/wailsapp/wails/v2/cmd/wails@latest). Development involves git clone, cd, and running ./dev.sh. Production builds use wails build.
• Prerequisites: API keys for supported AI providers (Gemini, OpenRouter, fal.ai, BytePlus) are necessary, configurable via environment variables, .env file, or application settings.
• Links: Homepage

Highlighted Details

• 8-Direction Sets: Generates 5 AI-driven directions and derives 3 mirrored directions, reducing generation costs by 37.5%.
• Self-Correcting Quality Loop: Employs up to 3 regeneration passes, guided by detected defects, to refine frame count and motion quality.
• Authentic Pixelization: Utilizes shared-palette quantization and pixel-grid snapping for true dot-art aesthetics.
• Engine-Friendly Export: Outputs sprite sheets, Aseprite-compatible JSON, manifest files, per-state GIFs/APNGs, and individual frame PNGs.
• Multi-Provider Support: Integrates with Gemini, OpenRouter, fal.ai, and BytePlus.
• Deterministic Correction: Leverages advanced techniques like YCbCr matting, DP optimal cut segmentation, and alpha-weighted centroid alignment for precise results.
• Quality Scoring: A 0-100 multi-axis score evaluates frame count accuracy, identity consistency, motion presence, and structural integrity.

Maintenance & Community

The project is maintained by "PerfectPixel contributors." No specific community channels (e.g., Discord, Slack) or roadmap details are provided in the README.

Licensing & Compatibility

• License: MIT.
• Compatibility: The MIT license is permissive, allowing for commercial use and integration into closed-source projects without significant restrictions.

Limitations & Caveats

Functionality is dependent on the availability and configuration of external AI provider API keys. Specific versions of Go (1.25+) and Node.js (18+) are required. The project focuses heavily on its pipeline's strengths, with no explicit mention of known bugs or unsupported platforms.