Turn a photo into pixel art you can refine.

Choose a photo with a readable subject

Pixel conversion removes information. Start with a photo whose important shapes still make sense when small: a face against a simple background, an object with a clear outline, or a landscape divided into strong foreground and background areas.

• Crop empty space before converting so useful pixels describe the subject.
• Prefer even lighting over tiny highlights and deep, noisy shadows.
• Use contrast between the subject and background to preserve the silhouette.
• Keep transparent PNG backgrounds when you are building an isolated asset.

Set Pixel Detail by what must survive

Pixel Detail is the number of editable cells along the photo’s longest edge. Ignore how large the preview looks on screen and inspect the actual grid. Increase Pixel Detail only until the defining details become readable.

If the face is technically recognizable but the expression is not, add a little detail. If the result looks like a reduced photograph rather than pixel art, lower Pixel Detail and simplify.

Reduce colors before adding texture

Begin with an automatic palette and no dithering. Lower the color count until skin, clothing, sky, or other major regions read as deliberate groups. A useful portrait starting point is 10–16 colors; a landscape can often support 14–20.

Add a small amount of diffuse dithering only when a smooth light transition matters. Ordered dithering is more visible and works best when you want the pattern itself to contribute to the style.

Clean clusters, not isolated mistakes

Zoom in and look for single pixels that interrupt a larger shape. Use the eyedropper and pencil to join those pixels to the nearest useful cluster. On portraits, give extra attention to the eyes, mouth, hairline, and the outer silhouette.

• Remove scattered colors from flat areas.
• Strengthen one-pixel breaks in an important outline.
• Use Restore when an edit should return to the generated base.
• Temporarily reveal the source image with O without tracing every detail.

Use Box, Lasso, or Color selection when a whole cluster needs to move or change. Copy a selection to a separate raster layer when you want to test a replacement without damaging the generated base. The canvas context menu can also move selected pixels to the editable layer immediately above or below.

The goal is not to reproduce every photographic edge. Preserve the few shapes that make the subject identifiable at the final viewing size.

Export for the destination

Choose native size when another pixel editor or game engine should receive the exact grid. Choose a whole-number scaled PNG for sharing on sites that would otherwise display a tiny image. Save a project file as well when you expect to revise the edit.

For a quick process post, choose Before / after comparison and drag the split between the visible pixel-art composite and the original photo. The result is one flattened PNG; it does not alter the editable project.