Pixels into print, symbols, and happy accidents.No prompts. Just a source and your decisions.
Drop an image to begin.
Dither Lab is a local, deterministic image-art instrument. Choose a source, reduce it to marks, and adjust until the texture feels right.
Choose a file, drop it anywhere, or paste an image. Browser-decodable PNG, JPEG, WebP, GIF, AVIF, BMP, and self-contained SVG work. Animated images become a still frame. HEIC and RAW are not universally supported.
Choose a process and palette — or let From the image derive one — then adjust scale, tone, and grain. Stack effect bakes your result into a new source for another pass. Undo restores the previous stage.
PNG, bit-packed PNG-8, WebP, JPEG, and AVIF export the rendered image. Smallest keeps whichever lossless file is smallest, and 1:1 exports a dither's bare sample grid, one pixel per cell. SVG uses actual shapes or text, not an embedded bitmap. ASCII and Braille can export plain text. Recipe files replay the complete effect stack on another source.
Processing changes the visual language of an image, not its provenance. An AI-generated source remains AI-originated. This tool makes no detector-evasion or human-authorship guarantees.
No network calls, analytics, remote fonts, AI models, or cloud storage. Images stay in this tab; save your image and recipe before closing. Working sources are limited to 3200 px on the long edge; exports to 2560 px. Complexity caps are shown when active.
A recipe stores the current controls, random seeds, and every baked stage. It does not contain the image. Load a source, then import a recipe to replay the entire stack from that original.
SVG exports also embed the recipe, including this note and the source filename. Recipes are data, not executable code. Reproduction can vary slightly between browsers because of image resampling and system fonts. Plain-text exports preserve symbols but not colors, jitter, or scanlines.