Photos contain more small color changes than most drawings. Hair, skin, leaves, shadows, camera noise, and JPEG compression all compete for a limited number of pixel cells. This guide is about preparing a photograph so the important subject stays recognizable after conversion, rather than about image conversion in general.
Decide what the pixel version must preserve
A portrait needs the face and pose to read. A product photo may need the silhouette and one defining detail. A landscape relies on large areas of light, horizon, and color. Pick the one feature that makes the image recognizable before moving a slider. That feature is the standard you can use when deciding whether the conversion worked.
A detailed group photo will rarely remain readable at 16 by 16 cells, even if the exported PNG is large. Each person needs enough grid space for a face or body shape. If the image has several subjects, choose a larger grid or split it into separate crops. The final cell count matters more than the final file dimensions.
Crop away background competition
A busy background consumes cells that could describe the subject. Use Crop Image to frame a person, animal, object, or scene feature more tightly. For portraits, keep the full outline of the head and any gesture that matters. Cropping too close can make hair, hands, or ears disappear when the image is reduced.
A square crop works well for avatars and square game assets. Fixed grid settings preserve the source proportions, including landscapes and wide compositions. Crop first when you want to change the framing, then check the original and result side by side before exporting.
Handle JPEG texture and compression
JPG files often contain fine texture and compression artifacts around high-contrast edges. Those marks can become unexpectedly prominent after downsampling. A cleaner, larger source usually gives the converter more reliable shapes than a tiny screenshot of a compressed social-media image. If you have the original photograph, use it instead of a repeatedly saved copy.
The converter accepts JPG as well as PNG and other image formats. It cannot recover details that compression already removed, and exporting as PNG afterward does not restore them. What it can do is reduce the image to a controlled grid and simplify colors. Inspect eyes, mouths, product edges, and text at the final viewing size, because these areas expose compression artifacts quickly.
Use a grid size that matches the subject
A 16 by 16 grid is a demanding choice for a real photograph. It can work for a close, high-contrast object, but most portraits benefit from 32 by 32 or 64 by 64 cells. Choose a fixed Pixel Grid Size to control the actual number of cells, then use Pixel Block Size to decide how large those cells appear in the exported image.
Start at 32 by 32 for a cropped head or a simple product. If the face or label becomes indistinct, move to 64 by 64. If the result looks like a low-resolution photograph instead of deliberate pixel art, try fewer colors or a smaller grid. There is no universal best setting: the most useful grid is the smallest one that keeps the chosen identifying feature.
Choose between faithful and stylized color
Original Colors is useful when the source's color is part of its identity, such as a pet's coat, a product package, or a familiar outfit. It can also preserve unwanted photographic noise. A limited palette compresses nearby shades into fewer choices and often makes larger areas easier to read, but skin, foliage, or a logo may change noticeably.
Dithering can suggest intermediate tones with a pattern of available colors. It may help a smooth sky or soft shadow, yet it can make a small face look speckled. Try no dithering first. Add an ordered or Floyd-Steinberg pattern only if a gradient truly needs it, then review at normal size. The pattern should support the image rather than become the most noticeable feature.
Finish for the destination
For an avatar or a social graphic, export a PNG and check it as a small thumbnail. For a craft pattern, enable the grid and export a PDF that can be printed. For code or an LED display, JSON gives cell-by-cell color data. These outputs share the same converted grid, but each is meant for a different next step.
A reliable photo to pixel art sequence is: choose the subject, crop it, select the actual grid size, compare color options, and inspect the downloaded file. When the background is still overpowering, return to the crop instead of immediately increasing resolution. A focused image usually gains more from better framing than from hundreds of extra cells.
Related guides
The practical takeaway
A photograph becomes useful pixel art when the subject gets enough cells and the background stops competing for them. Frame the subject first, then tune grid size, palette, and export for its final use.
Turn a photo into pixel art