How to compress an image without making it look bad

Resize before you compress

A 4,000-pixel-wide photo displayed in a 800-pixel column is carrying five times the detail anyone will see. Resizing to the size it is actually displayed at usually saves more than any quality setting.

For most web use, 1,200 to 1,600 pixels on the long edge is plenty. Thumbnails rarely need more than 600.

Choosing a quality setting

Between 70% and 80% is the sweet spot for photographs. Below 60% you start to see blocking in skies and gradients; above 90% the file grows quickly for detail nobody notices.

Screenshots and images with text or sharp edges are less forgiving. Keep them higher, or use PNG or lossless WebP.

Which format to use

WebP is typically 25 to 35% smaller than JPEG at equivalent visual quality and is supported by every current browser. JPEG remains the safest universal choice. PNG is for transparency and lossless copies, not photographs.

• Photographs for the web: WebP, JPEG as a fallback.
• Logos, icons and screenshots with text: PNG or lossless WebP.
• Anything needing transparency: PNG or WebP, never JPEG.

Avoid generation loss

Every lossy save discards detail permanently. Compressing an already-compressed file twice does visible damage, and the second pass often barely reduces the size.

Always compress from the original file. Keep the original somewhere safe and treat compressed copies as disposable output.

A note on privacy

Browser-based tools like ours process the file on your own device using a canvas element. Nothing is uploaded, which matters when the image is a document, an ID or anything you would not email to a stranger.