PNG Pixelate / Mosaic

What pixelating does

Pixelating an image divides it into a regular grid of squares and replaces every pixel inside each square with the average colour of that square. The result is a chunky, blocky version of the original — visibly made up of large squares of single colours. It's the look of early arcade graphics, Minecraft, and pretty much every "redacted face" you've seen on the news.

Mathematically, pixelating is a two-step process: shrink the image down to a tiny version (e.g. from 1920 wide to 80 wide for a 24-pixel block size), then scale that tiny version back up to the original dimensions using nearest-neighbour sampling so the edges stay crisp. The output canvas keeps the original dimensions but contains only as many distinct pixels as the shrunk middle version.

Why pixelation beats blur for privacy

Both blur and pixelation hide details, but only pixelation is genuinely hard to reverse. Standard Gaussian blur preserves more information than people realise — modern machine-learning techniques can recover surprisingly readable text and recognisable faces from heavily blurred images, especially when an attacker knows what they're looking for (a 16-character password, a numberplate, a specific person's face).

Pixelation at a sufficient block size destroys information rather than smearing it. Once the original colour values are averaged into a single square, the individual pixel-level variations are gone. There's no clever algorithm that can recover them. The trade-off is visible: pixelation is uglier than blur. But for redacting credit card numbers, passwords, ID documents, or any data with real consequences if revealed, pixelation is the right answer.

Rule of thumb: for serious privacy, the pixelation block size should be larger than the average size of a text character in the image. For a screenshot where text is ~12 px tall, use a block size of 16–24 px.

Creative uses for pixelation

• 8-bit / retro game aesthetic — channel the visual style of NES/Atari-era graphics. Combine with a limited colour palette for full retro effect.
• Mosaic art mock-ups — pixelating a portrait gives a quick preview of how it would look as actual tile mosaic.
• Censorship in television/media — pixelated faces and bodies in news/reality TV.
• Loading placeholders — heavily pixelated low-res versions of an image work as instant placeholders, similar to blurhash.
• Stylised social media posts — pixelated banners, retro-themed marketing.
• Glitch / vaporwave aesthetics — pixelation pairs naturally with other glitch effects.

Block size — what each range looks like

• 2–4 px — barely visible pixelation, more like aggressive resampling. Useful only as anti-aliasing for very sharp pixel art.
• 5–10 px — clearly chunky but original subject still recognisable. Soft mosaic feel.
• 11–18 px — strong pixelation, faces still vaguely identifiable, text scrambled to large unreadable shapes. Light-duty privacy redaction.
• 20–32 px — heavy pixelation, only the dominant colours and shapes remain. Serious privacy redaction starts here.
• 40–60 px — image becomes an abstract grid of large colour squares. Heavy stylisation.
• 60–80 px — only a few squares per image dimension. Pure abstraction.

How this tool works under the hood

Your PNG is drawn at a reduced size onto a temporary canvas. If the original is 1600 px wide and you've chosen a block size of 16 px, the temporary canvas is 100 px wide (1600 ÷ 16). That tiny image — where every pixel is now the averaged colour of one 16×16 block from the source — is then stretched back to 1600 px wide on a fresh canvas with imageSmoothingEnabled = false, so each "tiny pixel" becomes a perfectly sharp 16×16 square in the output. The result is exported as a fresh PNG.

The browser's built-in image scaling (used for the shrink step) averages multiple source pixels into each output pixel, which is exactly what we want — it's a fast, hardware-accelerated implementation of the mosaic average.

Tips and best practice

• For redacting credit card digits, passport numbers, license plates: use block size ≥ 20 px and prefer a solid black rectangle if absolute privacy matters.
• For redacting faces in social photos, block size 24–40 px is the journalistic norm.
• If your image is small to begin with (e.g. 400 px wide), drop the block size proportionally — 8 px on a 400 px image gives the same visual effect as 32 px on a 1600 px image.
• Don't combine pixelate with a blur — they cancel out the visual rhythm and produce a soft-but-blocky mess. Pick one.
• For "8-bit game" looks, follow up with Saturation bump to 130–150% to mimic the punchy colours of retro game palettes.

FAQs

Related tools