Comparison

Pixel art vs a pixelate filter

A pixelate filter makes a photo look blocky. A pixel-art grid replaces the photo with cells that each have one color. From across the room they can look similar. At 400% zoom they don't.

Filters usually average a block, then scale that block back up to the original pixel size with smoothing still partly in the pipeline. Edges pick up extra shades. You can't list the colors, because there are hundreds of near-grays along every border. You can't stitch it, because a "red" square is twelve reds. You can't drop it in a game as a 32×32 sprite, because the file is still 3000 pixels wide.

What you should be able to count

On Pixelate Tools the meta line is the contract. 48×36 means 48 columns and 36 rows. The small PNG's dimensions match that line. Open the file in any editor that shows image size. If it says 480×360, you downloaded the large preview, which is allowed, but it's a scaled view. The export page separates those on purpose. A filter app rarely offers the 48×36 file at all. It offers the big softened one, because that's what looks good in a story post.

Color is the second count. Set Colors to 8 and the legend should show eight swatches, each with one hex. A filter won't hand you that list. If you need the list for CSS or for thread, you need the grid. The pixel converter ends in that legend.

Where filters still win

A filter is faster when you don't care about the cells. You want a thumbnail to feel retro and you'll never edit a square. Use the phone. Come back when you need a sprite, a chart, or a file another program can treat as pixels. Doing both and comparing them once is worthwhile, so you stop expecting the filter's softness from a 16-color grid. The grid looks harsher because it threw information away on purpose.

AI is a third thing

Some sites sell "pixel art" by prompting a model. The model may draw a character that wasn't in your photo, change a logo's letters, or add shading that's not on a grid. Pixelate Tools doesn't do that. If you upload a logo, the logo's shape is what gets simplified. If the letters die, it's because 24 columns can't hold them, not because a model redrew the word. That predictability is the reason to use a converter. The definition page, what's pixel art, is the short version of this argument.

Try it on the Sunset example built into the tool. It's drawn in the browser, not fetched. Set width to 32 and colors to 6 with dithering off. You should see a sky, a sun, and a hill as flat regions. Then set dithering to Floyd–Steinberg. The regions break into mixed cells, still from those 6 colors. A filter wouldn't be able to tell you the six hex values. The legend can. If you can't see squares, zoom the small PNG, not a screenshot of the page.

The same test catches a resized preview. If the file's pixel width is 320 and the meta line said 32, you're looking at the large PNG, and a filter-like softness can appear when a chat app scales that file down again. Download the small PNG, which will be 32 pixels wide, and zoom inside an editor that has a nearest-neighbor option. Squares that stay squares are the grid. Squares that grow gray edges were resampled by the viewer. That's the viewer's problem, and it's also why the small file exists.

When a filter is the better choice

Use a filter when you need a one-off mood image, when you'll never edit individual cells, and when no game engine or craft chart consumes the file. Use the converter when the downstream step counts cells, colors, or symbols.

Three-question grid test

Compare file weight: an 8-color 32×32 PNG is often a few kilobytes. A filter output at 2000 pixels wide is hundreds of kilobytes of near-duplicate shades. Size on disk isn't art quality, but it's a hint about whether the pipeline reduced colors.

Questions

Keep going

• Image to pixel art
• What is pixel art?
• Small vs large PNG
• Image to pixel art

Updated 30 September 2026