WebP vs JPEG compression ratio: the short answer
WebP is usually the smaller file at equal visual quality, but the ratio is a per-image result, not a constant.
- Direction: WebP produces fewer bytes than JPEG for photographic images at matched visual quality.
- Size of the gap: the ratio moves with the quality setting, the amount of detail in the image, and how hard the source JPEG was already compressed.
- When it flips: heavily compressed JPEG sources, flat graphics and text, and images with transparency can make the WebP file equal or larger.
- How to compare: match the rendered images first, then count bytes. The quality numbers in the two formats do not line up.
- What to standardize: one tested setting per image class — photos, screenshots, graphics — not one setting for a whole site.
What the compression ratio means
The compression ratio compares input bytes with output bytes: a 600 KB JPEG that becomes a 400 KB WebP has a ratio of 1.5:1, or a 33% saving. Three conditions make that number meaningful.
- Same content and dimensions. Cropping or resizing changes the comparison.
- Same visual quality. The two files must look equivalent, not share a quality number.
- Same delivery path. A file that is re-encoded in transit is not the file you measured.
Change any one condition and the ratio answers a different question.
Why WebP usually wins on photos
WebP encodes photographs differently from JPEG, and the difference favors smooth, detailed images.
JPEG splits an image into 8×8 blocks, transforms each block, quantizes the result, and codes the survivors with a Huffman table.

WebP predicts each block from its neighbors, transforms the prediction error instead, and codes the output with an adaptive entropy coder that updates its probabilities as it works.
Two effects follow. Prediction removes the cost of repeating smooth areas, and adaptive coding spends fewer bits on symbols that appear often. Together they let WebP reach a given level of visible quality with fewer bytes — which is exactly what a compression ratio measures.
Where the ratio narrows or reverses
WebP does not win everywhere. The gain shrinks in predictable situations.
The source JPEG is already heavily compressed. A file saved at a low JPEG quality has little redundancy left. Re-encoding it removes more detail without freeing many bytes, so the WebP can match the original size or exceed it.
The image is mostly flat color or text. Screenshots, line art, and diagrams carry little photographic detail. A lossless PNG or a lossless WebP often beats both lossy formats here, and lossy WebP can add ringing around sharp edges that costs bytes and looks worse.
Transparency is involved. WebP carries an alpha channel; JPEG cannot. The extra channel adds bytes, so compare a transparent WebP against a transparent PNG, not against an opaque JPEG.
The quality setting goes high. Near the top of the range both formats spend bytes freely, and the gap between them narrows.
The comparison uses the quality number instead of the eye. WebP quality 70 is not JPEG quality 70. The scales come from different encoders, so matching numbers produces a matched-numbers comparison, not a matched-quality one.
How to measure the ratio on your own images
The only ratio that matters is the one your images produce. The loop below takes a few minutes per image class.
- Sample the class. Pick 5 to 10 images that represent it: a hero photo, a detail shot, a flat graphic, and something with texture or noise. One image is not a sample.
- Keep the dimensions. Encode at the size you will ship. Never compare a resized file with an original.
- Match the quality by eye. Encode the JPEG and the WebP, then adjust the WebP setting until the two are hard to tell apart on the details that matter.
- Record the bytes. Divide the JPEG size by the WebP size for the ratio. Subtract the result from 1 for the percentage saved.
- Decide per class. If the WebP is smaller and looks equal, adopt that setting for the class. If it is larger, keep JPEG for that class.
- Batch last. Apply the chosen setting to the full set only after the sample passes.
Compare at 100% zoom first, then step back to normal viewing size. If the WebP bands a smooth sky or smears fine texture, raise the quality setting before you count bytes.

Two criteria decide the outcome: the WebP file is smaller, and the image still looks equal at normal size. If either one fails, the conversion is not a win for that image class.

Tuning the setting that drives the ratio
The quality value is the control that moves the ratio the most. Each step down the scale is a trade-off: fewer bytes against more visible damage. A few rules keep the tuning honest.
- Start from the encoder default and move in steps of five to ten points.
- Test on a copy. Overwrite nothing until the setting is settled.
- Change one variable at a time: quality first, then encoder effort, then metadata handling.
- Re-test after an encoder upgrade. Defaults and coding tools change between versions.
Higher encoder effort — the method setting in most WebP tools — usually buys a smaller file at the same quality in exchange for encoding time. It does not change how the image looks.
Applying the ratio check on a Mac
The loop above works with any encoder. Tools differ in how much of it they automate.
Image2WebP is a Mac utility from HP60 Studio, which publishes this site.
HP60 Studio's Image2WebP lists batch conversion of JPEG and PNG files on the Mac, and its Pro tier is described as preserving EXIF and GPS metadata during conversion, so the file details a photographer relies on survive the round trip.
cwebp, the reference encoder from the WebP project, exposes quality and method controls from the command line, which suits scripted or one-off conversions.
Match the tool to the job. A folder of client photos favors a queue you can leave running on the desktop. A single file or a build step favors the command line.
Follow-up questions
The questions below are grouped by what a reader wants to do next.
Measuring and comparing
What is the WebP-to-JPEG compression ratio? There is no single value. Compare both formats at the same visual quality and the same dimensions, then divide the JPEG size by the WebP size. Photographs usually show the largest gap; flat graphics show the smallest.
Is WebP always smaller than JPEG? No. WebP can match or exceed the JPEG size when the source was already compressed hard, when the image is mostly flat color, or when transparency is added.
Can a website calculator give me the answer? A calculator can demonstrate the idea on one demo image, but your ratio depends on your content and your settings. Measure your own sample instead of trusting a generic number.
Settings and quality
Does WebP quality 70 equal JPEG quality 70? No. The scales come from different encoders and do not map one to one. Match quality by comparing the rendered images, then record the WebP value that works for that image class.
Should I switch to lossless WebP? Use lossless when the output must be pixel-exact: screenshots, line art, text, interface captures. For photographs, lossy WebP usually reaches equal visual quality at a much smaller size.
Does lossless WebP beat PNG? Both are lossless, so compare them on size and on tooling. WebP often packs flat graphics tightly, while PNG remains the more universal format for archival and interchange. Measure your own files before standardizing.
Other formats
WebP or AVIF for a photo-heavy site? AVIF is the newer format and can produce smaller files at the same visual quality, at the cost of slower encoding and less mature tooling. WebP encodes faster and has broader support. Test both on your own sample if your pipeline handles AVIF.
Where does JPEG XL fit? JPEG XL is a separate format with its own encoder and its own support picture. Compare it with the same method: match visual quality first, then compare bytes. Do not judge it by its quality number.
Should I replace PNG with WebP everywhere? Replace PNG where a lossless WebP is smaller and your audience's browsers accept WebP. Keep PNG where a downstream tool or an archival requirement expects it.
Workflow
Can I convert a whole folder at once? Yes, after the sample passes. Batch conversion saves time, but it also applies one setting to every file, so tune first and batch second.
Do I need to keep the originals? Keep them until the converted set has been reviewed. The ratio tells you what you saved; it does not tell you whether a detail mattered.


