A quality value such as 80 is not a universal scientific percentage. Different encoders map their sliders to different quantization choices, color handling, and optimization strategies. A JPEG saved at ‘80’ in one application can differ in both appearance and file size from an ‘80’ saved elsewhere.
What the quality control changes
JPEG divides image information into frequency components and stores them with reduced precision. Lower quality generally discards more subtle information, especially in fine texture and color detail. The savings are not linear: moving from 95 to 85 can cut many bytes with little visible change, while moving from 30 to 20 may create obvious damage for a smaller additional saving.
Four artifacts to watch for
- Block boundaries in smooth gradients or skies.
- Ringing or halos around text, logos, and high-contrast edges.
- Smearing in hair, foliage, fabric, or other fine texture.
- Color bleeding around saturated red, blue, or purple details.
Choose quality by destination, not by a magic number
| Use case | Priority | What to inspect |
|---|---|---|
| Product image | Texture and edge clarity | Labels, stitching, surface finish |
| Profile photo | Faces | Eyes, hair, glasses, skin gradients |
| Screenshot | Text and UI edges | Small fonts, icons, thin rules |
| Email photo | Reasonable size | Normal viewing quality rather than pixel-peeping |
If the destination gives you a maximum file size instead of a quality number, it is usually more useful to target the file size directly. Fitabyte Compress searches for settings that stay below the requested maximum, then you inspect whether the visual result is still fit for purpose.
For controlled examples of JPEG and WebP sizes on different image types, see the Fitabyte Lab. Those tests show why the same quality setting produces very different byte counts for a flat graphic and a photo-like scene.