Guide
How to Compress JPG Files: Quality Settings Explained
JPG compression works in two stages: a mathematical transform (DCT) splits the image into 8×8 blocks, and a quantization step throws away high-frequency detail. The quality setting controls how aggressive that discard is.
Which quality number to use
- 90-100: near-lossless. Rarely worth the file size for web use.
- 75-85: the sweet spot. Most photos look identical to the original at normal viewing size.
- 50-70: visible but usually acceptable artifacts on fine textures.
- Below 50: blocking and color banding become obvious.
Hitting an exact size
Upload forms often demand "under 100KB". Guessing with a slider is slow. A better tool re-encodes the image several times, binary-searching the quality until the file fits - typically converging in 4-6 attempts.
One warning
JPG is lossy every time. Compress once from the original, not from an already-compressed copy, to avoid stacking artifacts.