Advice such as ‘WEBP is smaller’ or ‘quality 80 is enough’ is common, but it rarely shows the actual difference when one source is saved at the same dimensions. We encoded one Toolit brand image as WEBP and JPG at four quality levels, plus one PNG re-encode.
This experiment measures file size only. It does not represent perceived quality or every kind of photograph. The source is a single graphic containing text, icons, and gradients, and that limitation matters when interpreting the numbers.
Three distinctions to make first
At every tested quality number, the WEBP result was smaller than JPG.
WEBP grew by about 2.7× from quality 80 to 95.
Re-encoding this image as PNG produced a file larger than the original.
How to check it yourself
- 1
The source was Toolit’s 1,741,024-byte, 1536×1024 PNG containing text, icons, and gradients.
- 2
In Chrome, we opened Manual settings in Toolit’s image compressor and set the longest side to 1920px. The source is smaller, so every output remained 1536×1024.
- 3
We saved WEBP and JPG outputs at quality 40, 60, 80, and 95 and recorded the displayed result size.
- 4
PNG is lossless and ignores the quality setting, so it was re-encoded once and measured separately.
- 5
Each setting change triggered a new compression run. Values use the one-decimal KB figures displayed by the tool.
Measured results
All outputs below kept the same 1536×1024 dimensions. The original PNG was about 1.7MB.
| Format | Quality | Result size | Displayed reduction |
|---|---|---|---|
| WEBP | 40 | 37.1KB | 98% |
| WEBP | 60 | 45.6KB | 97% |
| WEBP | 80 | 64.8KB | 96% |
| WEBP | 95 | 172.9KB | 90% |
| JPG | 40 | 62.4KB | 96% |
| JPG | 60 | 79.2KB | 95% |
| JPG | 80 | 115.2KB | 93% |
| JPG | 95 | 278.8KB | 84% |
| PNG | Not applicable | 2.3MB | Larger than source |
The jump from quality 80 to 95 was unusually large
WEBP increased from 64.8KB at quality 80 to 172.9KB at 95, about 2.7 times larger. JPG increased from 115.2KB to 278.8KB, about 2.4 times larger. A 15-point quality increase does not imply a 15% size increase.
If upload size matters, quality 80 is a practical place to begin and inspect the result. Graphics with small text or sharp edges may still justify a higher setting after visual review.
This does not prove that WEBP is always better than JPG
For this source, WEBP was roughly 38–44% smaller than JPG at the same quality number. That does not mean WEBP 80 and JPG 80 have identical perceived quality: quality scales and encoders are not directly equivalent across formats.
The source is a UI graphic rather than a photograph. Portraits, night scenes, and highly textured photos can produce different gaps. Test a few representative images from the real workflow before standardizing a setting.
Why the re-encoded PNG was larger
PNG stores pixels losslessly instead of accepting a quality setting. If the original was already optimized with a different PNG encoder, a browser Canvas re-encode can be larger. Here, the approximately 1.7MB source became 2.3MB.
Choosing PNG does not automatically reduce file size. When transparency or lossless pixel preservation is unnecessary, WEBP or JPG is more likely to create a smaller output.
Limits of this table
Do not apply one image’s results to every photo. Browser and encoder versions, texture, colors, transparency, metadata, and output dimensions all affect the result.
- Results reflect Chrome and the Toolit implementation tested on August 20, 2026.
- We set numeric quality values but did not run PSNR, SSIM, or a human visual-quality study.
- The source was smaller than 1920px, so every output remained 1536×1024.
- Canvas re-encoding does not retain location and camera metadata in the output.
Verifiable claims matter more
For this graphic, WEBP quality 80 produced a 64.8KB file and was a useful starting balance, while quality 95 jumped to 172.9KB. JPG was larger at every measured step but remains a practical choice when the destination requires broad compatibility.
The durable lesson is not to memorize one format or quality number. Test representative images and decide whether target size, compatibility, transparency, or fine-detail preservation matters most.