WASM
Two-layer cleanup for document photos: remove the handwriting first, then blend faint white or gray residue back into the paper background
Loading the local processing engine...
Auto mode focuses on colored handwriting. Switch to dark mode for thicker dark pen marks.
Fill Method
Higher sensitivity removes more aggressively. Dark mode can also affect bold printed text.
Multi-layer cleanup
Layer 1: Main handwriting removal
Keep the current handwriting detection and local inpainting pass to clear the visible strokes first.
Layer 2: Subtle trace cleanup
Run a second background-fusion pass over faint white or gray residue so the cleaned area blends back into the paper texture.
Higher values erase faint residue more aggressively, but can also smooth nearby printed texture.
Reference photo (Optional)
If you still have the original photo with colored markup, upload it too so the remover can infer a stronger handwriting mask.
✓ Per-pixel background estimation handles non-uniform lighting accurately
✓ Dual-channel detection: color deviation + luminance drop captures all handwritten ink
✓ Distance-weighted inpainting blends filled regions naturally with surroundings
✓ Second-pass residual blending suppresses faint white or gray ghosts so the paper background stays coherent
✓ Runs the pixel pipeline locally in Rust/WASM for fast browser-side processing