Quality
How to compress images without losing quality
2 min readUpdated 29 September 2026By QANEN
“Without losing quality” means one of two things, and knowing which you want is the whole game. Lossless compression keeps every pixel identical to the original. Visually lossless compression throws away data your eyes are very bad at noticing, so the image looks the same even though it no longer is, byte for byte. You can get dramatically smaller files either way.
Lossless vs visually lossless
Lossless — the decoded image is mathematically identical to the source. Best for logos, screenshots, line art, product shots on flat backgrounds, and anything you might edit again. Savings are real but modest compared with lossy.
Visually lossless — a carefully tuned lossy encode where artefacts are hidden in busy detail the eye glosses over. Best for photographs. This is where the big size reductions live.
Where the quality actually goes
Lossy image formats (JPEG, WebP, AVIF) shrink files by approximating small blocks of colour. Push the quality too low and you get visible problems:
- Blocking — visible squares in smooth areas like skies.
- Ringing / mosquito noise — fuzzy haloes around sharp edges such as text or wire fences.
- Colour blotching — subtle gradients turn into bands.
The skill is finding the lowest quality setting before those appear, not the smallest file at any cost.
Settings that work in practice
Quality scales are not comparable across formats or even tools, so treat these as starting points on a typical 0–100 scale:
| Goal | Format | Starting point |
|---|---|---|
| Photo, website, general use | WebP | Quality ~80–85 |
| Photo, smallest modern file | AVIF | Quality ~60–70 |
| Aggressive but still clean | WebP / AVIF | Quality ~65, then eyeball it |
| Must be pixel-perfect | Lossless PNG / WebP | Lossless (quality 100) |
Swipe the table sideways to compare every column →
In Image Compressor by Qanen these map to presets: Balanced (quality 82) for everyday photos, Web Ultra (quality 65) when you want the smallest web file you can still defend, and Lossless for pixel-perfect output.
The workflow that protects quality
- Start from the highest-quality source you have. Never compress an already-compressed, artefacty image — each lossy pass compounds damage. Ideally work from the original.
- Resize before you compress. A 6000 px photo shrunk to the 1200 px you’ll actually display is the single biggest saving available — and it costs no perceived quality. (Resizing is separate from compression; see reducing image size for a faster website.)
- Pick the right format for the content. Photos → WebP/AVIF; graphics that must stay exact → lossless PNG. (See WebP vs AVIF vs JPEG.)
- Compare at 100%. Zoom to actual pixels and check edges, faces and text — not just the thumbnail.
- Only re-compress from originals. Keep your uncompressed masters so you never degrade generation after generation.
The fastest way to internalise this is to try it: open the compressor, load a photo, and drag the before/after slider at different quality values until you can’t tell them apart. That’s your personal visually-lossless point.
Common myths
- “80% quality means it keeps 80% of the quality.” No — the number is an encoder-specific knob, not a percentage of fidelity.
- “Lossless always looks better.” A well-tuned visually-lossless photo can be indistinguishable from lossless at a fraction of the size.
- “Converting JPEG → PNG makes it lossless again.” It changes the container, not the pixels; the damage from the original JPEG is already baked in.