Every image compressor confronts you with the same fork in the road: lossy or lossless? One can shrink a photo by 90%, the other guarantees perfection. Understanding what each actually does to your pixels lets you choose confidently instead of guessing.
Lossless: perfect, but heavy
Lossless compression (PNG, lossless WebP, and formats like ZIP) works by finding redundancy — repeated patterns, runs of identical pixels, statistical regularities — and encoding them more efficiently. Nothing is thrown away; decompressing reproduces the original bit-for-bit.
That works brilliantly on images with large uniform areas: screenshots, logos, diagrams, flat-color illustrations. A screenshot full of white background and repeated interface elements compresses beautifully without losing a pixel.
It works poorly on photographs. Real-world photos are full of sensor noise and subtle variation — almost no two adjacent pixels are identical — so there's little redundancy to exploit. A photo saved as PNG is routinely 5–10× larger than a JPEG that looks identical to the eye.
Lossy: exploiting the limits of human vision
Lossy compression (JPEG, lossy WebP) takes a different bet: most of the data in a photograph is detail your visual system cannot perceive anyway, so why store it? It discards information in a carefully prioritized order:
- Fine color detail goes first. Human vision resolves brightness far more sharply than color, so color information is stored at reduced resolution — usually imperceptibly.
- High-frequency texture goes next. Subtle grain in a patch of grass or sky is approximated rather than recorded exactly.
- Only at aggressive settings does it touch detail you'd actually notice — which is when the familiar "blocky" artifacts appear.
The quality slider controls how far down this list the encoder goes. At quality 75–85, a photo typically shrinks by 70–90% while remaining visually indistinguishable from the original at normal viewing size.
The catch: lossy is one-way and cumulative
Two properties of lossy compression deserve respect. First, it's irreversible — discarded detail cannot be recovered, so always keep originals of anything irreplaceable. Second, it's cumulative — every re-save of a JPEG recompresses it, and artifacts build up over generations of editing. Edit from originals; compress once, as the final step before publishing or sharing.
How to choose, in practice
- Photograph → lossy. JPEG or WebP at quality 70–85. This is what lossy compression was invented for.
- Screenshot, logo, diagram, text → lossless. PNG or lossless WebP. Lossy artifacts smear exactly the sharp edges these images depend on, often without even saving much space.
- Archival master copies → lossless (or the camera's original file). Compress derivatives, never masters.
- Not sure? Compress lossy at quality 80 and compare before/after at full size. If you can't see a difference — and for photos you almost never will — take the smaller file.
The bottom line
Lossless preserves everything and pays for it in size; lossy discards what you can't see and wins enormously on photos. The right answer isn't one or the other — it's matching the method to the image. Photos travel lossy, graphics travel lossless, and originals stay safely uncompressed at home.
Compress your images now
Ready to put this into practice? Use the free WebPcompress compressor — it runs entirely in your browser, handles 20 images at a time, and your photos never leave your device.