Compression Methodology
How Canvas rendering, quality search, target tolerance, resolution reduction, PNG handling, color, and metadata work.
Canvas rendering
The browser decodes the source and draws it to Canvas or OffscreenCanvas at the calculated output dimensions. This creates a new raster rather than modifying the source file.
Quality search
JPEG and WebP target-size mode tests encoder quality with a bounded search. The selected result is the highest tested quality below the limit when such a result exists.
Target-size tolerance
The default tolerance is used to avoid meaningless repeated attempts around encoder steps. The tool reports the actual bytes rather than claiming exact equality.
- Keep the original file before editing.
- Check dimensions, crop, transparency, and file size after download.
- Do not send private source images when reporting a problem.
Resolution reduction
If minimum useful quality remains too large and automatic reduction is allowed, the dimensions are reduced gradually with a minimum output boundary.
PNG behavior
Normal browser PNG export is lossless and does not expose a comparable lossy quality control. PNG may require smaller dimensions or conversion to another format.
Color, metadata, and browser differences
Canvas export normally removes metadata and may normalize color through the browser’s sRGB-oriented pipeline. Codec and color results can vary across browsers and operating systems.
Use the browser-based editor
Apply exact dimensions, target-size compression, and batch settings to your own files.
Open Image Resizer