2 September 2026

Resize first,
compress second

Most people reach for the quality slider and leave the dimensions alone. That is the wrong order, and it costs visible quality for very little saving.

The arithmetic nobody does

A recent phone camera produces images around 4000 pixels along the long edge. A full-screen laptop display is about 1500 pixels wide. An image in an email or a document is rarely shown wider than 1000. File size scales with pixel count, which scales with the square of the dimension, so halving the longest edge removes roughly three quarters of the data before compression is touched at all. That is the largest single saving available, and it is invisible because nothing was being shown at full size anyway.

Where the quality slider shows first

JPEG compression works on eight-by-eight blocks and discards fine detail before coarse detail. That means degradation appears first at hard edges: text, logos, and sharp boundaries against flat colour. A landscape photograph still looks clean at 60% because there are few hard edges in it. A screenshot or a document scan at 60% looks obviously damaged. The right setting depends on what is in the picture, not on a general rule.

Why re-compressing an already compressed image hurts twice

Every JPEG save is lossy, and the losses accumulate. Opening a photograph, saving it, opening it again and saving it again produces visible artefacts even if the quality setting never changes, because the second encode is working on the first encode's approximation rather than the original. This is why a tool that copies compressed data across untouched is better than one that decodes and re-encodes it, wherever the operation permits.

What compression quietly removes

Because compressing means decoding the picture and encoding a new one, nothing from the original file's metadata survives. The output has no GPS coordinates, no camera serial number, no timestamps and no owner name. People often compress an image for size and receive a privacy improvement they did not ask for and do not know they have. It is worth knowing which way round that works: compression removes metadata, but converting formats frequently preserves it.

A reasonable default

For a photograph going into an email, a document or a website: 2000 pixels on the longest edge, JPEG, 80% quality. That is typically around a tenth of the original file with no difference anyone will see. Start there and only move if the picture contains text or fine line work.

Try it