Pixelate Image

Pixelate Image provides deterministic, non-destructive pixelation for photos and graphics. It is intended for journalists, researchers, designers, and privacy officers who must redact faces, license plates, or other identifiers before sharing media. Use occurs when quick, repeatable pixelation is required without altering non-redacted content.

The tool supports a block-based pixelation algorithm with user-controlled resolution. Required features include image upload, pixel_size_px input (4–64), and region selection (entire image or defined bounding boxes). Optional/advanced features include per-region pixel sizes, batch processing via multi-file input, output format selection, and preservation of transparency for PNG. Results are deterministic given inputs and parameters.

Inputs comprise image_file (required; JPEG/PNG/WEBP), pixel_size_px (integer, 4–64), and regions (optional list of bounding boxes). Outputs include a pixelated_image in the chosen format, with metadata noting pixel_size_px and regions used. The tool preserves non-target content and returns a new image file rather than modifying the source.

The core method partitions the target region into non-overlapping blocks of size pixel_size_px × pixel_size_px. Each block is recolored to the block’s representative color (average color) using a deterministic, per-block calculation, then upsampled via nearest-neighbor to form a blocky mosaic. Multiple regions are processed independently; coordinates are clipped to image bounds.