Generate production-ready sprites from text prompts. Background removal, edge decontamination, and pixel-perfect grid reconstruction — all via a single API call.
curl -X POST https://genr.dev/api/sprites/generate \
-H "x-api-key: genr_your_key_here" \
-H "Content-Type: application/json" \
-d '{"prompt": "a fire-breathing dragon"}'
# Returns: { image: "base64...", passed: true, processingTimeMs: 4200 }A battle-tested pipeline that combines AI generation, computer vision, and pixel art algorithms for production-quality output.
Describe a sprite in plain text, or upload your own image. Our Gemini-powered prompt engineering uses magenta keying for perfect transparency.
Magenta color-keying + BRIA RMBG neural network. Two passes catch what one misses — 97% quality pass rate.
Edge-guided grid detection via autocorrelation. One clean color per cell, mathematically perfect pixel art with binary alpha.
BFS distance mapping detects JPEG compression halos. Contaminated pixels are color-corrected or removed. Multi-pass until clean.
97% pass rate on 66 test sprites including transparency stress tests — doughnuts, glass bottles, crystal balls, stained glass windows.
Start free. Scale when you need to.
10 sprites/month
200 sprites/month
1,000 sprites/month