Most Images Are Far Larger Than They Need to Be
A photograph straight off a modern phone runs to four or five megabytes, and almost none of that is needed for a website, an email attachment or an upload form with a size cap. The detail is real, but at the size the image will actually be displayed, most of it is invisible.
This compressor lets you trade quality for size and see the result before committing, so you can find the point where the file stops shrinking usefully and starts looking worse.
How to Compress an Image
The quality slider is the main control, and the right value depends on what the picture contains.
- Drop your images onto the box above. You can process several at once.
- Set the quality. Around 80 is the usual starting point for photographs, which typically cuts the file to a fifth of its size with no visible difference at normal viewing size.
- Compare the before and after sizes shown for each file.
- Inspect the result at full size before accepting it, particularly on faces, skies and any area of smooth gradient, where artefacts appear first.
- If the file is still too large, consider reducing the pixel dimensions as well, an image displayed at 800 pixels wide gains nothing from being 4,000 pixels wide.
- Download the compressed versions; the files on your device are left exactly as they were.
Quality Settings Are Not Percentages
A quality setting of 80 does not mean the image keeps 80 per cent of its information, and 50 is not half as good as 100. The number is an index into how aggressively the encoder quantises detail, and the relationship with both file size and visible quality is non-linear. Most of the size reduction happens between 100 and 80, while most of the visible damage happens below about 60.
That shape is why 80 is such a common recommendation: it sits after the steep part of the size curve and before the steep part of the damage curve. Going from 100 to 80 might save 70 per cent of the file; going from 80 to 60 saves much less and costs considerably more.
Quality Settings in Practice
Approximate behaviour for a typical photograph. Content varies these figures considerably.
| Quality | Size vs original | Visible difference | Use for |
|---|---|---|---|
| 100 | ~90% | None | Archival, further editing |
| 90 | ~40% | None at normal size | Print, high-detail display |
| 80 | ~20% | None at normal size | The general default for web |
| 70 | ~15% | Slight on gradients | Thumbnails, dense galleries |
| 60 | ~12% | Visible on skies and faces | When size is the priority |
| 40 | ~8% | Clearly degraded | Rarely worth it |
Notice that the saving between 80 and 60 is small while the damage is substantial. That asymmetry is the reason to resist pushing quality down when a file is still too large, resizing the image instead usually gives more with less.
Resizing Beats Compressing
Halving the width and height of an image quarters the pixel count, which typically does more for the file size than any quality setting, and costs nothing at all if the image was larger than its display size to begin with. An image shown at 800 pixels wide on a page is being scaled down by the browser anyway, so the extra pixels were only ever download weight.
Compression is also generational. Each round of lossy encoding discards information the previous round kept, so repeatedly compressing an already-compressed file degrades it further even at high quality settings. Always work from the original rather than from a copy that has been through the process before.