Pixel Lab: Turn Any Image into Pixel Art,
Dither, ASCII and Sprite Sheets
Drop a photo, an illustration, a GIF or a video clip and Pixel Lab rebuilds it as clean pixel art: 15 dither modes, palette control, cleanup, texture grammars and an FX lab, then export as SVG, PNG, sprite sheet, GIF, WebM or animated SVG. It runs in your browser, nothing is uploaded, and it's free.
Controls on the left, the canvas in the middle, quality score and output metadata on the right. Thirty production recipes sit above the canvas for a one-click start.
What Pixel Lab Does
One tool, 22 modules. Most people only need the first five, so Pixel Lab opens with the essentials and lets you switch the rest on when a job calls for them.
Image to pixel art
Source, auto setup, resolution and framing. Drop any PNG, JPG, WebP, SVG, GIF or video, set a pixel budget (25K to 400K cells) or a fixed grid, choose cover or contain, and the auto setup picks sampling, palette and dither for you. Change one slider and the preview follows live.
Dithering and colour
Floyd-Steinberg, Atkinson, Stucki, Burkes, Sierra, Sierra Lite, Jarvis-Judice-Ninke, random threshold, blue-noise, clustered dot, checker and sparse transitions, and ordered Bayer 2x2, 4x4 and 8x8, with a strength slider and serpentine diffusion. Palettes come from the image, from presets (including the GetIllustrations brand palette) or from your own list, with a minimum separation so colours never muddy.
Cleanup and reconstruction
Edge protection, one-pixel worm removal, island removal, stair-step regularisation and neighbour cleanup passes turn a noisy conversion into something that looks drawn. Adaptive reconstruction keeps fine detail where the image needs it and simplifies where it doesn't.
SVG and PNG export
Master SVG with one square per cell, an optimised SVG that merges same-colour cells into larger rectangles, pixel forms (square, circle, diamond, beveled, dot), and PNG at 1x to 16x. A GPU preview renders a 4K frame when you need print size.
Sprite sheets, GIF, WebM, animated SVG
Feed it a video or a folder of frames and get a packed sprite sheet with JSON metadata, a GIF or WebM at your frame rate, or an animated SVG that plays in any browser without a player.
ASCII, textures and FX
Seven ASCII renderers plus Braille and Matrix modes, a texture grammar (engraving, crosshatch, weave, facet, dissolve), procedural texture overlays and an FX lab with CRT curvature, glitch, bloom, chromatic aberration, film grain and more. Stack effects and randomise them.
AI masks and zones
A SegFormer model runs in your browser to separate subject from background, so you can dither only the sky or stylise only the figure. Paint your own mask on top. Optional, free, and the model downloads once.
Bulk and collections
Folder rules, saved presets, batch export of a whole folder to SVG, PNG and JSON, and a collection style system that fingerprints up to 10 examples and keeps a set of images consistent. Made for people converting a pack, not a single file.
How to Use Pixel Lab
Six steps from a source image to a finished export. The first three are enough for most jobs.
Drop your source
Drag an image, GIF, video or folder onto the canvas, or paste from the clipboard. Nothing leaves your machine.
Let auto setup run
Pick a quality preset and a pixel budget. Auto tune reads the image and sets sampling, palette size and dither to a sensible first pass.
Tune palette and dither
Lock colours you want to keep, swap the dither algorithm, drag the strength slider and watch the preview update as you go.
Clean it up
Remove stray pixels and islands, protect edges, regularise stair-steps. The quality score on the right tells you what changed.
Export
SVG, PNG, ASCII text, sprite sheet ZIP, GIF, WebM or animated SVG. Settings and palettes export as JSON so a teammate can reproduce the look.
Save the recipe
Save your settings as a preset, then apply it to a folder with the bulk module and every image in a set comes out matching.
Dither and palette, live
Module 06 holds the dither mode, strength, serpentine diffusion and pattern scale; module 05 the palette. Every change re-renders the preview, and the Pixel preview tab shows the raw cells before any texture or FX. The style deck above the canvas applies a whole recipe in one click: Pink Dusk Seascape, Ember Forest Poster, Mist Lake Halftone and 27 more.
Switch modules on when you need them
Sprite sheet, GIF and video, adaptive reconstruction, tone shaping, semantic zones, pattern lab, collection style, GPU preview, FX lab and procedural texture each have a switch. Off, they cost nothing. On, they add their controls to the left rail and their outputs to the export list. The FX preview tab here is running the ASCII renderer.
Export Formats
Everything renders locally and downloads straight from the page.
| Format | What you get | Best for |
|---|---|---|
| Master SVG | One rectangle per cell, every pixel editable in Figma or Illustrator | Editing the artwork further, scaling without blur |
| Optimised SVG | Same-colour cells merged into larger rectangles, far smaller file | Websites, icons, anything that ships |
| PNG | 1x to 16x scale, transparent or flattened background | Social, docs, game assets |
| Sprite sheet ZIP | Packed tiles plus JSON frame metadata | Game engines, CSS sprite animation |
| GIF and WebM | Frame rate, duration, scale and bitrate you choose | Previews, social posts, READMEs |
| Animated SVG | A self-contained SVG animation of the FX cycle | Logos and marks that have to stay vector |
| ASCII text | Plain-text art in seven character sets, Braille or Matrix | Terminals, code comments, GitHub profiles |
| Settings and palette JSON | Every slider and colour, reproducible | Teams, batch jobs, version control |
What People Make With It
A few starting points, with packs from the library that suit each one.
Game sprites and sheets
Turn concept art or a short clip into a packed sprite sheet with frame data. Start from the Robot Pixel Art pack if you need characters that are already pixel-native.
Retro brand marks
An 8-bit or CRT version of a logo for a launch page or a sticker. Export the optimised SVG so it stays crisp at any size, and read our guide to vector pixel art in Illustrator for the finishing steps.
Halftone and engraved print
Clustered-dot and halftone screens with the engraving grammar give photos a printed feel. The Halftone Doodle characters show the look drawn by hand.
ASCII and terminal art
README banners, CLI splash screens, code-comment art. The Claude Mascot pack is a popular source for this one.
Animated pixel logos
Run the FX cycle on a mark and export an animated SVG or a GIF. Glitch, bloom and CRT presets are tuned for short loops.
Consistent sets
Converting a whole illustration pack? Fingerprint a few examples, lock the master palette and run the folder through the bulk module so every piece matches.
Under the Hood
Pixel Lab is a single-page app written in plain JavaScript with no framework and no build step. Sampling, palette quantisation, dithering, cleanup and the SVG optimiser all run on the main thread against canvas pixel data; the 4K preview and the FX lab render on the GPU through WebGL. Video and GIF sources are decoded frame by frame in the browser, and the sprite packer, GIF encoder and WebM recorder are in-page too.
The one thing it fetches is the optional AI model. The semantic zones module loads Transformers.js and a SegFormer checkpoint (about 40 MB) from a public CDN the first time you press Load AI model, then keeps it cached. If you never open that module, the page makes no network requests after it loads.
Your settings, saved presets and collection fingerprints live in this browser's local storage. Export them as JSON if you want to move them or share them. The code is open source under the MIT licence: github.com/RamyWafaa/gridlab has the full technical documentation, module by module.
Frequently Asked Questions
Short answers about the tool, privacy and what you can do with the output.
Is Pixel Lab free to use?
Yes. Pixel Lab is free, there is no account to create and no watermark on exports. It runs entirely in your browser, so there is nothing to install either.
Does my image get uploaded anywhere?
No. Every conversion happens on your own machine. The only thing the tool ever downloads is the optional AI segmentation model, and that is fetched from a public CDN, not from us. Your files never leave your computer.
What can I turn into pixel art?
PNG, JPG, WebP and SVG images, GIF and video frames, or a whole folder at once through the bulk module. Drop the file on the canvas, pick a pixel budget and the auto setup does the first pass for you.
Which dithering algorithms are included?
Fifteen: Floyd-Steinberg, Atkinson, Stucki, Burkes, Sierra, Sierra Lite, Jarvis-Judice-Ninke, random threshold, a blue-noise style, clustered dot, checker and sparse transitions, and ordered Bayer at 2x2, 4x4 and 8x8. Dither strength is a slider, serpentine diffusion is a switch, and the FX lab adds mono and CMYK halftone screens on top.
Can I export pixel art as SVG?
Yes, two ways. The master SVG keeps one square per cell so every pixel stays editable, and the optimised SVG merges neighbouring cells of the same colour into larger rectangles for a much smaller file. PNG export goes from 1x up to 16x, and the GPU preview renders a 4K frame.
How do I make a sprite sheet?
Load a video clip or a set of frames, open the Sprite sheet module, choose the tile size, columns and padding, then export. You get the sheet as a ZIP with a JSON file describing every frame, ready for game engines and CSS animations.
Can it make GIFs or animated SVGs?
Yes. The GIF / Video module exports GIF and WebM at the frame rate, duration and scale you choose, and the animated SVG export turns the FX cycle into a self-contained SVG animation that plays anywhere an SVG does.
What is ASCII mode?
A set of text renderers that redraw your image with characters instead of pixels: standard, minimal, dense, technical, hatching, terminal blocks, Braille dots and a Matrix digit rain. The result is plain text you can paste into a README, a terminal or a code comment.
What does the AI mask do, and does it cost anything?
The AI semantic zones module runs a SegFormer model in your browser through Transformers.js to separate subject from background, so you can dither or stylise only part of an image. It is free. The model is about 40 MB and downloads once from a public CDN the first time you load it.
Can I use what I make commercially?
Anything you create from your own images is yours to use however you like. Pixel-art versions of GetIllustrations packs follow the licence of the pack they came from. The tool itself is open source under the MIT licence on GitHub.
More Free Design Tools
Browser-based, no installation, made by the same team.
SVG Color Editor
Recolour any SVG in the browser: every fill, stroke and gradient detected, live preview, preset palettes.
Open the editor →Font Pairing Tool
Preview curated Google Fonts pairings with live loading and copy the CSS.
Open the font tool →AI Studio
Draft a one-off illustration in the art direction of a pack you already own. For members.
Open AI Studio →