Image compressor

Resizing usually beats compressing

A 4000-pixel-wide photo displayed in a 800-pixel column is carrying 25 times more pixels than anything on screen will use. Setting the longest edge to what the image is actually displayed at will usually save more than any quality setting, and — unlike lowering quality — it costs nothing visible.

Do both: resize first, then compress what is left.

What the quality slider does

For JPG and WebP, quality controls how aggressively detail is discarded. The useful range is narrower than the slider suggests:

• 0.8–0.9 — visually indistinguishable from the original for most photos.
• 0.7–0.8 — the usual choice for the web. Artefacts appear only under close inspection, at half or less of the file size.
• Below 0.6 — visible blocking around edges and in flat areas such as sky.

Photographs compress well because their detail is irregular. Screenshots, diagrams and anything containing text do not: JPG smears text edges badly, and those images belong in PNG or WebP.

Compression is not reversible

Every re-encode discards information permanently. Compressing an already compressed JPG at quality 0.8 does not give you an image at 0.8 — it gives you one that has been through the process twice, and the artefacts compound. Always compress from the original.

The canvas also drops metadata. That deletes the EXIF block, which removes GPS coordinates and camera details — often welcome — but also the orientation flag and any embedded colour profile, so a wide-gamut photo can come out looking flat. Keep the original for anything you intend to edit or print.

Your files stay here

Images are decoded and re-encoded in your browser using Canvas. Nothing is uploaded, so there is no size limit imposed by a server and no copy of your photos on someone else’s disk.