How to Turn a Photo Into Pixel Art Using Photoshop

How do you turn a photo into pixel art in Photoshop?

Boost contrast, crush the color palette with Indexed Color, then downscale and upscale using Nearest Neighbor resampling. That's the whole method. No pixelate filter, no mosaic filter, no plugin required. Pixel art is digital art built pixel by pixel, from a deliberately limited palette with no anti-aliasing — a flat definition, but a load-bearing one. This walkthrough covers the real manual version, done inside Photoshop itself. Want the tool-agnostic version of the same idea? See the companion guide. Everything below is something you click yourself, so along the way you'll also see exactly where kott automates the palette math in one pass.

Why does Nearest Neighbor matter more than any filter?

Adobe defines Nearest Neighbor as a fast, lower-precision resample method that replicates existing pixels rather than averaging them — built for artwork with hard, non-anti-aliased edges. That single sentence is the whole trick. Bicubic, Bicubic Smoother, Automatic (which defaults to Preserve Details on a photo) — every other resample option in Photoshop exists to blend neighboring pixels for a smoother, more photographic result. Which is exactly backwards for a pixel art conversion. You need the opposite at both ends: when the photo shrinks down to a handful of pixels, and again when those pixels get blown back up to something readable.

Skip Nearest Neighbor and nothing else you did matters. Crush the palette to eight colors, and you'll still land on soft, blurred blocks — the resample step just reintroduced the smoothing you spent two prior steps removing.

What are the exact steps, start to finish?

1. Open the photo. JPG or PNG, doesn't matter — you're about to throw away most of the detail anyway.
2. Image > Adjustments > Brightness/Contrast. Push contrast up — +25 is a solid starting point on most photos. It sharpens tonal edges before you quantize color, so the palette reduction in the next step gets cleaner boundaries instead of a soft gradient to work with.
3. Image > Mode > Indexed Color. Set the palette somewhere around 8-16 colors. This is the step that actually makes it read as pixel art instead of a low-res photo — a forced, limited palette is what your eye interprets as "art" rather than "compression artifact."
4. Image > Image Size, with Resample turned on. Drop the height drastically — around 130px tall for a typical photo — and set the Resample dropdown to Nearest Neighbor. This downscale is what actually creates your pixel grid: every pixel in the shrunk image is a straight sample, not a blend.
5. Image > Image Size again. Scale back up by an integer multiple — 400% is typical — with Resample still set to Nearest Neighbor. This is what makes each of those tiny pixels visible again as a clean, square block instead of a blurry patch.

That's the entire method. No pixelate filter step, no mosaic filter, no third-party plugin — contrast, palette, and two Nearest Neighbor resizes in sequence.

Worth being straight about what this manual method doesn't give you. Automated converters like photo2pixel expose separate controls for kernel size (how much neighboring color blends during downsampling), output pixel size, and an edge threshold that draws outline pixels at detected edges — three independent knobs, tuned per image. Photoshop's manual version collapses all of that into one blunt instrument: the contrast slider in step 2. Reasonable as a stand-in for edge sharpening, but it's no substitute for real edge detection. On a busy or low-contrast source photo, you'll notice the difference — fine detail either survives as noise or disappears entirely, with no middle setting to reach for.

Why do default Photoshop settings make pixel art look fake?

Two failure modes cover almost every "why doesn't my pixel art look right" question, and both trace back to skipping a setting above.

First: resample choice. Bicubic-family methods and Preserve Details average neighboring pixels, which produces halos around shapes in high-contrast areas and soft, blurred edges once the image gets enlarged again. Left Resample on default and only did the palette reduction? This is almost certainly what happened.

Second: palette size. Push Indexed Color past roughly 16-20 colors and the image stops reading as a deliberately limited palette — it just looks like a compressed photo. A hard palette limit only works because it's visibly a choice, not a byproduct. For the version of this problem that shows up as pattern noise instead of blur, there's the dithering half of Photoshop's manual pixel-art toolkit. Dithering and flat Indexed Color solve different problems — know which one you're actually reaching for.

Nearest Neighbor vs Photoshop's other resample methods

One row matters more than the rest: the first one. Every other option exists to make photos look smoother — precisely the thing you're trying to undo here.

One practical wrinkle. Scaling by a non-round percentage can still introduce soft pixels even with Nearest Neighbor selected, because Photoshop has to decide how to handle fractional pixel boundaries somewhere. Stick to integer multiples — 200%, 300%, 400% — for both the downscale and the upscale. Need a final size that isn't a clean multiple? Resize to the nearest integer multiple first with Nearest Neighbor, then take one last pass down to your exact target dimension using Automatic interpolation. That last pass reintroduces a bit of blending, but only once, on an image already committed to its pixel grid. The result holds up better than trying to hit an odd percentage in a single step.

Frequently Asked Questions

Do I need a pixelate or mosaic filter to make pixel art in Photoshop?

No. A contrast boost, an Indexed Color palette cut, and a Nearest Neighbor downscale/upscale do the whole job. A mosaic filter blurs block edges together instead of sharpening them, which is the opposite of what you want.

Why does my pixel art look blurry after resizing it in Photoshop?

The Image Size dialog defaulted to Bicubic or Preserve Details, which averages neighboring pixels to smooth the result. Reopen Image Size and set Resample to Nearest Neighbor before scaling in either direction.

How many colors should I use for a pixel art conversion?

Somewhere around 8-16 colors via Image > Mode > Indexed Color is the usual working range. Fewer than 8 starts losing shading detail; more than 16 stops reading as a limited palette and looks like a filtered photo instead of pixel art.

Can I get an exact pixel dimension that isn't a round percentage?

Yes. Upsample first with Nearest Neighbor to get comfortably past your target size, then downsample to the exact dimension using Automatic interpolation for the cleanest final result.

Conclusion

Three settings, in order: contrast, Indexed Color, Nearest Neighbor on both resize steps. Everything else in Photoshop's Image Size and Indexed Color dialogs is built for photographs, not pixel grids — leave it on default and it will quietly work against you. Rather skip the manual palette math? Or want to try the same conversion on a photo you don't want to touch destructively? Try kott's pixel effect with a preset already loaded, and compare the result against your own Photoshop pass. Photoshop gives you finer per-region control if you're touching up a single frame by hand; kott gets you there faster across a whole batch. Pick based on which one you're actually doing.