What is being compared?
The lab starts with one source and exports it into three formats with your browser’s encoder. It measures the generated files, rather than showing estimated savings or a fixed ranking. PNG ignores the lossy-quality setting. JPEG cannot keep transparency: the selected background fills transparent areas. WebP support is checked against the actual exported file type.
This is a bounded comparison, not a full-resolution archival converter. The input limit is 10 MB and 12 megapixels; processing uses at most a 1600-pixel long edge and does not enlarge a smaller source. Matching dimensions keep comparisons useful, but reducing dimensions is itself a change. Keep the original file. Maximum input side: 16,384 pixels.
Three samples with different failure points
The gradient illustration makes banding and block artifacts easier to spot. It is deliberately synthetic, so its results are not evidence about all camera photographs. The small-text and line sample exposes ringing or softened edges: a smaller file can still be a poor choice if its labels become difficult to read.
The transparent badge helps you check what a format preserves. Compare its outline against the checkerboard, then change the JPEG background. A white rectangle in the JPEG is expected; it is not recovered transparency. PNG and WebP can retain the alpha channel, although lossy WebP may still change color details.
A repeatable comparison, not a universal score
For a controlled comparison, keep the source, output dimensions, quality value and JPEG background unchanged while comparing formats. When you alter a setting, run again and compare the newly generated files. Smaller bytes alone do not establish better quality, and there is no single quality setting that is equivalent across different encoders.
The source bytes describe the chosen original file; output bytes describe the generated exports. An already optimized input may be smaller than every result. Browser version, encoder implementation and sample text rendering can affect the numbers. Record your settings and keep the downloaded files if you need to reproduce or challenge a result.
Choose the next edit for the destination
If a form accepts only JPG, choose a readable JPG even when WebP is smaller. For a logo that needs a transparent edge, compare PNG and WebP in the intended background and browser. For a screenshot containing fine text, inspect readability before lowering quality further.
Use the image resizer or cropper when dimensions or framing are the problem. Use target-size compression when an upload has a byte limit, and confirm that the actual output is below that limit. This lab compares formats; it does not promise a target size, lossless conversion to JPEG, or a quality score.
Frequently asked questions
Why do my results differ from another device?
Browsers can use different encoders and text rendering. Compare exports generated together in the same browser, using the same source and settings.
Does the lab upload my picture?
No. The application processes it in browser memory. Downloads are created locally; reloading clears the lab state.
Why is a converted file bigger?
The source may already be efficiently encoded. Re-encoding, keeping transparency or changing format can increase bytes. Keep the original if the new file offers no useful benefit.