Lossy vs lossless — when each makes sense

Every image compression conversation eventually hits the same fork: lossy or lossless? The answer depends on what you're compressing and where it's going.

What's the difference?

Lossless compression reduces file size by finding patterns in the data. No information is thrown away — the decompressed image is bit-for-bit identical to the original.

Lossy compression throws away information the human eye is unlikely to notice. The result is smaller, but the original data cannot be reconstructed.

Think of it this way: lossless is a ZIP file for images. Lossy is an aggressive summariser that keeps the gist but drops the footnotes.

When to use lossy

Lossy compression is the right choice for:

• Photographs — the complex colour gradients and noise in photos compress extremely well with lossy algorithms
• Hero images on websites — file size directly affects page load time
• Social media assets — platforms re-compress anyway, so pixel-perfect doesn't matter
• Email images — smaller is always better for email deliverability

Recommended formats: AVIF, WebP, JPEG

Quality sweet spot: 76–80 for web delivery, 85–90 for source files you'll edit later

When to use lossless

Lossless compression is the right choice for:

• Screenshots with text — lossy compression creates visible artefacts around sharp text edges
• UI assets (icons, buttons, logos) — these have hard edges and solid colours that look muddy with lossy compression
• Medical or scientific images — accuracy matters more than file size
• Source files you plan to edit later — every lossy re-save degrades quality further (generation loss)

Recommended formats: PNG, lossless WebP, TIFF (for archival)

The quality ladder

Here's what happens as you decrease JPEG quality from 100 to 10:

| Quality | Visible effect | | -------- | ---------------------------------------------------------------------------------------- | | 100–90 | No visible difference. File is 2–3× larger than necessary. | | 85–76 | Still imperceptible on web. The sweet spot for most uses. | | 75–60 | Subtle softening visible at 1:1 zoom on retina. Usually fine at normal viewing distance. | | 59–40 | Obvious softening. Colour banding in gradients. | | 39–20 | Heavy artefacts. Blocky 8×8 patterns visible in JPEG. | | Below 20 | Barely recognisable. Useful only for thumbnails. |

WebP and AVIF degrade more gracefully than JPEG at equivalent quality numbers. An AVIF at quality 50 typically looks better than a JPEG at quality 70.

The generation loss problem

Each time you open a JPEG, edit it, and save it again, quality degrades. This is called generation loss. After 5–10 round-trips, the degradation becomes visible.

The fix: always keep a lossless master copy (PNG or TIFF). Edit from the master. Export lossy only for final delivery.

Practical decision tree

1. Is it a photograph or complex image? → Lossy (AVIF > WebP > JPEG)
2. Does it have sharp text or hard edges? → Lossless (PNG or lossless WebP)
3. Will you edit it again later? → Keep a lossless master, export lossy for delivery
4. Is file size irrelevant (local storage, archival)? → Lossless always

Try it yourself

Drop an image into our compressor and compare the output at quality 90 vs quality 78. You'll see the file size drop by 30–40% with no visible quality loss. That's the power of understanding where your images sit on the lossy-lossless spectrum.