How to reduce image size without losing quality
Updated
"Without losing quality" is not quite achievable in the literal sense — every lossy compression discards something. What is achievable, and what people actually mean, is losing nothing you can see at the size the image will be viewed.
The trick is the order of operations. Most people reach straight for a quality slider, which is the least effective lever and the most damaging. Cropping and resizing first removes far more bytes for zero visible cost, leaving compression with an easy job instead of an impossible one.
Step by step
Crop away anything you do not need
This is free. Pixels you remove cost nothing and improve the composition. On a photographed document, cropping the surrounding desk can halve the file before anything else happens.
Resize to the dimensions it will actually be displayed at
This is the big one. An image shown 800px wide on a page gains nothing from being 4000px wide — but it costs 25 times the pixels. Halving both dimensions removes 75% of the data with no visible change at display size.
Only then apply compression
With sensible dimensions, quality 80–85 is visually identical to the original for photographs. It is only when dimensions are far too large that the encoder is forced into visible artefacts.
Choose the right format
WebP typically produces files 25–35% smaller than JPEG at matching quality, and supports transparency. Use JPEG for maximum compatibility, PNG only for screenshots and line art.
Things worth knowing
- Enlarging cannot add detail. Upscaling a small image interpolates from neighbouring pixels and always looks softer than a photograph genuinely captured at that size.
- Re-compressing an already-compressed JPEG stacks artefacts. Always work from the original.
- Screenshots and diagrams belong in PNG or WebP. JPEG puts visible halos around sharp text edges.
- Re-encoding strips EXIF, which removes GPS coordinates — useful before sharing photos publicly.