What happens between choosing your image and downloading the result.
1. Check and decode
We inspect file signatures, declared types, extensions, and container dimensions. Files are limited to 20 MiB, 32 megapixels, and 12,000 pixels per side to reduce memory pressure. The browser decodes the image and applies its embedded orientation.
2. Resize and encode
The image is drawn on a white canvas. Fit retains the complete image with padding when necessary; Fill crops the edges centrally. JPEG quality is sampled within a bounded range, then refined around a promising candidate. Auto dimensions use a preferred quality floor and reduce resolution when necessary. Fixed dimensions allow stronger compression and show a warning when it is needed. Quality parameters are not a guarantee of perceptual quality or a globally optimal result.
3. Verify the result
We check the real byte count, JPEG format, decodability, and output dimensions. Only a verified result becomes downloadable. Automatic resizing can reduce dimensions to avoid heavy compression; explicitly supplied dimensions remain fixed.
Know what changes
Transparency becomes white. Animation is not preserved; animated inputs are processed as a still frame. Re-encoding does not preserve most original EXIF, GPS, or ICC metadata and may change colors. Check color-critical work in a dedicated editor. Processing varies with your browser and device.