Pixel Art Converter
Convert a photo or illustration into a grid of flat pixel blocks. The tool averages the visible colors inside each source block, then optionally maps that color to a limited palette. Compare it with the original and save a PNG with crisp, uninterpolated block edges.
Key features
- Choose source blocks from 1 to 512px, including partial right and bottom blocks
- Use uniform RGB palettes of 8, 27 or 64 colors, 4 or 16 grays, or the tool’s own 16-color palette
- Keep unrestricted block colors when only pixelation is needed
- Average color with alpha weighting and keep block transparency
- Export at original dimensions or scale each grid cell by an integer from 1 to 16
- Inspect grid size, logical RGB color count and output PNG dimensions
How to use
- Choose a static PNG, JPEG or WebP, or load the colored example.
- Set the source block size. Larger values produce a coarser grid.
- Choose a fixed palette or keep the averaged block colors unrestricted.
- Keep source dimensions or select an integer scale for the logical grid.
- Create the pixel art, compare it and download the current PNG. Generate again after changing settings.
Use cases
- Create a coarse pixel version of a profile illustration
- Explore limited colors for a retro-style graphic
- Prepare a small logical grid and enlarge it without soft edges
- Compare the same image with RGB and grayscale palettes
Frequently asked questions
How is this different from resizing or adding a photo filter?
It first groups the source into blocks and computes one alpha-weighted color per block. It then maps those colors to an optional fixed palette and expands the resulting grid without interpolation. Changing image dimensions or brightness alone does not perform those steps.
How are the limited palettes chosen?
RGB 8, 27 and 64 use respectively 2, 3 and 4 evenly spaced levels per channel. Grayscale uses 4 or 16 evenly spaced levels from the mean of the RGB channels. The custom 16-color option picks the closest entry in this tool’s fixed table using RGB distance. These palettes are fixed, not extracted from the image, and no dithering is applied.
What do block size and integer output scale mean?
Block size is measured in source pixels. The grid dimensions are the source dimensions divided by that size and rounded up. Integer scale sets how many output pixels each grid cell occupies. For the 800×520 example with 16px blocks, the grid is 50×33; at scale 4 its output is 200×132px.
Are pixels at the right and bottom edges discarded?
No. A final partial block uses all remaining source pixels and its actual sample count. Source-size output keeps that last block narrower or shorter. Integer-scale output makes every grid cell equally square, so the partial edge cells become full-size cells and the overall aspect ratio may change slightly.
How are transparency and color counts handled?
Block RGB is weighted by alpha, so invisible colors do not create dark fringes. Block alpha is the average source coverage. The reported color count covers visible logical block RGB values before PNG encoding and excludes alpha differences. Transparency and canvas rounding can affect colors in the encoded or composited image; the PNG is not an indexed-palette file.
What are the limits and what happens when I cancel?
Input supports one static PNG, JPEG or WebP up to 20 MiB, 20 million pixels and 8,192px per side. The logical grid is limited to 262,144 cells; use larger blocks when needed. Output is limited to 20 million pixels and 8,192px per side. Cancelling discards old results, and an already-running PNG encoder must finish cleanup before another job can start.
Privacy
Images and conversion settings stay in the current page’s browser memory. This tool does not upload them or automatically store them. Downloading saves a newly encoded PNG on your device; original file bytes, EXIF and embedded profiles are not preserved.
Comments & questions