JPEG Quality Settings Explained (and What Numbers to Use)
What JPEG quality values actually control, why quality 80 is not "80 percent", and the ranges that work for photos, screenshots and print.
The quality slider in an image editor looks like a percentage. It is not. Understanding what the number really does makes it much easier to pick one and stop worrying.
What the number controls
JPEG splits an image into 8×8 blocks, converts each to frequency coefficients, then divides those coefficients by values from a quantization table. High-frequency coefficients (fine detail, noise, sharp edges) get divided by larger numbers and end up mostly zero, which is what makes the file small.
The quality setting selects how aggressive that table is. It scales a table defined by the JPEG standard in 1992. There is no unit, no percentage, and no guarantee that two encoders agree: quality 80 in one tool may produce the same visual result as quality 85 in another, at a different file size.
Ranges that work
For photographic images at full resolution:
- 95–100 — archiving. Huge files, differences invisible outside a magnified crop.
- 90 — high quality. Good for anything that will be edited or printed.
- 80–85 — the web sweet spot. Most people cannot tell it from the original at normal viewing size.
- 70–75 — acceptable for large images displayed small, and for backgrounds.
- 50–65 — visible softening, but often fine for thumbnails and previews.
- under 40 — blocky, especially in smooth gradients like skies.
Screenshots and synthetic images behave differently: they contain hard edges and flat colour, which JPEG handles poorly at any quality. If your source is a screenshot, consider PNG for storage, or expect to stay at q90 or above.
Progressive or baseline?
A baseline JPEG stores its data top to bottom. A progressive JPEG stores several passes of increasing detail, so the image appears blurry-then-sharp while loading. For the same visual quality, progressive encoding is usually 2–8% smaller, and it looks better on slow connections.
This tool always writes progressive JPEGs.
The trap: re-compressing an already-compressed file
Every JPEG encode loses information. Compressing a q95 photo to q80 costs you something. Compressing that q80 result to q80 again costs you more and saves almost nothing, because the encoder is now spending bits describing the artifacts of the previous pass.
Our rule is simple: if re-encoding would produce a file larger than your input, we keep your input and tell you. And if your file is already compressed, we try a lossless pass first — rewriting only the entropy coding, leaving the image data bit-for-bit identical. On typical baseline JPEGs that saves about 4%, at zero quality cost.
Choosing quality vs choosing a size
If you have a hard limit — a form that rejects anything over 2 MB, a marketplace that wants thumbnails under 200 KB — do not guess the quality. Switch to target size mode, enter the limit, and we search for the highest quality that fits. That takes a few extra seconds and removes the guesswork entirely.