JPG to Pixel Art Converter

Most of the world’s photos are JPGs — camera shots, phone pics, downloaded images, social media saves. JPG is a lossy format, which means compression artifacts (blockiness, ringing around edges, banding in skies) hide in every file. Pixelation amplifies those artifacts if you don’t tame them first. This JPG-to-pixel-art page is built around that reality: the cleanup settings that neutralize JPEG noise before conversion, so your photos come out crisp instead of crunchy.

Why JPGs need a cleanup pass

JPEG compression works by discarding subtle detail in 8×8 blocks — invisible at full size, but pixelation is essentially a magnifying glass for block structure. Convert a JPG naively and you get exaggerated block edges, halos around high-contrast edges, and color banding where the compressor quantized a gradient. The fix isn’t complicated: the converter’s downsampling stage uses smooth (medium-quality) resampling, which already averages away fine compression noise, and the settings below finish the job.

Artifact-taming settings

• Edge Sharpening, used with restraint (15–30%): defines real edges without amplifying artifact edges. Cranking it to 100% on a JPG will sharpen the compression blocks themselves — don’t.
• Floyd–Steinberg dithering (70–90%): error diffusion is the best weapon against banding — it breaks up the stepped gradients JPEG leaves in skies and skin into smooth, natural-looking texture.
• Perceptual CIELAB matching: keeps the slightly-off colors JPEG introduces from drifting further during quantization.
• Moderate pixel blocks (6–12px): blocks much smaller than the JPEG’s own 8×8 artifact grid can resolve the artifacts as “detail”; slightly larger blocks average them away.

Recommended settings for JPG photos

For a typical phone photo: pixel size 6–8px, Floyd–Steinberg at 85%, CIELAB distance, Full Color / Adaptive palette at 16 colors, sharpness 25%, contrast +15. Heavily compressed images (old memes, multi-saved social images) benefit from larger blocks (10–14px) and a smaller adaptive palette (8–12 colors) — fewer, bolder colors hide a multitude of sins. High-quality camera JPGs can be treated almost like PNGs: smaller blocks, more colors, crisper results. The live split-slider preview is your truth serum — drag settings and watch the artifacts appear or vanish in real time.

JPG vs PNG as a source: honest comparison

For graphics, logos, and screenshots, PNG wins outright — lossless means nothing to clean up. For photographs, the gap is small: a high-quality JPG (camera “fine” mode, minimal recompression) converts nearly as well as a PNG once the cleanup settings are applied. The danger zone is repeatedly recompressed JPGs — images saved and re-saved across social platforms accumulate generational artifacts. If your source looks crunchy at 100% zoom before conversion, expect to use larger blocks and stronger dithering. When you have the choice, upload the PNG; when you only have the JPG, this page’s settings close 90% of the gap.

Portraits, pets, and landscapes

JPG is the native format of phone photography, so most portrait conversions start here. Faces follow the same rules as the photo to pixel art guide: 4–8px blocks, dithering on for skin tones, SNES or Adaptive palettes. Pet photos with fur texture do well with a touch more sharpening (30–35%) to keep whiskers defined through the JPEG softness. Landscapes with big skies are the classic banding victims — that’s where Floyd–Steinberg earns its keep. Whatever the subject, export doesn’t inherit the input format: your JPG becomes a crisp PNG (up to 16×) or vector SVG, with zero JPEG artifacts in the output file.

Social-media JPGs: the recompression ladder

Every time a JPG is uploaded, downloaded, and re-uploaded, it climbs the recompression ladder — each generation bakes in more artifacts. A meme that’s done the rounds on three platforms can be surprisingly crunchy underneath. For these veterans: larger blocks (12–16px), smaller palettes (8–12 colors), and Bayer dithering instead of Floyd–Steinberg — the ordered pattern masks block structure better than error diffusion on heavily damaged sources. Paradoxically, these abused images often make the most characterful pixel art: the damage becomes texture, and the result looks authentically lo-fi rather than accidentally broken.

Knowing when a source is beyond saving

Honesty matters: some JPGs are too far gone. Warning signs — large flat areas of banding that dithering can’t break up, text burned into the image at tiny sizes (it becomes unreadable dashes), and subjects smaller than ~10% of the frame (there simply aren’t enough pixels of the subject to pixelate meaningfully). In these cases the kindest advice is a better source: re-download the original, screenshot at higher resolution, or crop tighter before uploading. The converter is powerful, but it can’t invent detail that compression destroyed — and we’d rather you know that before spending twenty minutes on settings.