There is a persistent myth that image compression always visibly degrades image quality. This is true for bad compression. Good compression — using the right algorithm, at the right quality setting, in the right format — is visually indistinguishable from the original, even at 60–80% smaller file sizes.
The difference between the two is understanding what's happening under the hood.
How lossy compression works (and why it doesn't always mean quality loss)
Lossy compression works by discarding image data that the human visual system is least likely to notice. The science behind this — psychovisual modelling — has been studied extensively since the 1970s.
The key insight: your eyes are significantly better at detecting luminance (brightness) differences than chrominance (colour) differences. Lossy compression exploits this by preserving brightness data in high detail while compressing colour data more aggressively — in areas of an image where the two values are spatially close, you won't notice.
Modern formats like AVIF take this further, using more sophisticated perceptual models derived from video codec research. At the same subjective quality level, AVIF discards less perceptually important data and more efficiently encodes what remains — which is why it achieves smaller files at equal visual quality compared to JPEG.
The quality setting: where the real difference lies
Most compression tools expose a quality slider (usually 0–100). The choice of quality level is the single biggest variable in the quality/size trade-off.
| Quality range | Visual result | Best for |
|---|---|---|
| 90–100% | Virtually lossless | Source files, print, archiving |
| 80–90% | Indistinguishable on screen | Client delivery, high-quality web |
| 70–80% | Excellent for most uses | Web, email, social media |
| 60–70% | Minor artefacts visible on close inspection | Thumbnails, previews |
| Below 60% | Visible degradation | Not recommended for final output |
For most web and email use, 75–85% quality in JPEG or WebP, or 70–80% in AVIF, produces results that are indistinguishable from the original at normal viewing sizes. The file size reduction at these settings is typically 60–80%.
Format matters as much as quality setting
Choosing the right format is at least as impactful as the quality slider. Two images at identical subjective quality can have dramatically different file sizes depending on format:
- A photograph at perceptually identical quality: JPEG 420KB → WebP 210KB → AVIF 130KB
- A logo or icon: PNG (lossless) → WebP lossless (25% smaller) → SVG (if vector source exists)
The practical takeaway: for photographs and product images, use AVIF or WebP. For logos and UI elements requiring pixel-perfect clarity, use PNG or WebP lossless. Never use PNG for photos — the file size penalty is enormous with no visual benefit.
Strip metadata
JPEG files often contain significant metadata embedded by cameras: GPS coordinates, camera model, lens information, colour profiles, and more. This metadata is invisible in the image but can add 50–200KB to a file.
Good compression tools strip unnecessary metadata automatically during compression. TinyPixels removes EXIF metadata by default while preserving colour profiles — the metadata that actually affects how an image renders.
Avoid re-compressing already-compressed images
One of the most common mistakes: taking a JPEG (already lossy), editing it, and re-saving as JPEG. Each save cycle introduces a new round of compression artefacts, compounding on top of previous ones. After several cycles, quality degrades significantly even at high quality settings.
The correct workflow:
- Keep your working files in a lossless format (PNG, TIFF, or the camera's RAW format).
- Compress to the delivery format (JPEG, WebP, AVIF) once, as the final step.
- Never re-export from a previously compressed JPEG if quality matters.
The right tool for the job
For individual files, macOS Preview and online tools are adequate. For batch compression of dozens or thousands of files — which is the reality for designers, photographers, and developers — you need a dedicated tool with:
- Batch processing (drag a folder, get a folder back)
- Configurable quality per format
- Output to a separate folder (never overwrite originals)
- No upload requirements — and Pro removes all file size and batch limits
TinyPixels is built around this workflow. Drop a folder of mixed JPEG, PNG, and WebP files; set your target format and quality; get a folder of compressed output files back — entirely on your machine, in seconds. The originals stay untouched.
The visual test
If you can tell the difference between the original and compressed version while looking at a normal screen at a normal viewing distance, the quality setting is too low. Zoom in to 200%+ to see compression artefacts — but judge quality at 100% actual size, because that's how your users will see it.