Guide

How to compress images without it showing

Compressing means losing information; the craft is losing only what the eye cannot see. Here is the method, with numbers.

“Compress without losing quality” is, strictly speaking, impossible in JPG and WebP: compressing is losing information. But the question people are really asking has a good answer: how to remove as much weight as possible without the loss showing. You can, and by far more than you would think.

First, the setting that saves the most: dimensions

A typical phone photo is 4032 × 3024 pixels: twelve million pixels. If you are going to show it in an article 800 pixels wide, the browser will throw away 96% of those pixels when painting it. It downloaded them anyway.

So the right order is resize first, compress second. Dropping the longest side from 4032 to 1600 pixels usually removes 85% of the weight without touching compression quality, and without anyone noticing anything. Only after that does the quality slider deserve attention.

Second, the quality level: where the curve bends

The relationship between quality and size is not linear. Going from 100 to 90 removes a lot of weight and almost no visible quality. From 90 to 80 is still very profitable. From 80 to 70 becomes arguable. Below 60 the classic artefacts show up: 8 × 8 pixel blocks in skies, halos around edges and a dirty texture across flat colours.

Third, the right format

Changing format can save more than any setting. A photographic PNG converted to JPG at quality 85 drops 80–90% of its weight. That same JPG converted to WebP drops another 25–30%. If the destination is a website, converting to WebP is the biggest lever you have.

The mistakes that ruin an image

Re-compressing over and over

Every time you save a JPG it is compressed again on top of an already compressed image, and the damage accumulates. Ten save-and-resave cycles leave a photo visibly dirty. Always work from the original and produce the compressed version once, at the end.

Enlarging a small image

Pixels that do not exist cannot be invented. Blowing a 400 px image up to 1600 px only produces a blurrier, heavier version. If you need more resolution, you need the original, not a resizer.

Using JPG for screenshots

Text and fine lines are exactly what JPG handles worst. A screenshot at quality 75 comes out with dirt around every letter. For screenshots, always PNG (or lossless WebP).

Compressing the same file twice “just in case”

It improves nothing and degrades a little more. If the result does not get smaller, the file was already optimised.

How to check whether it shows

The professional trick: open the original and the compressed version in two tabs at 100% zoom and flip between them with Ctrl+Tab. If something “jumps” when you switch — usually skies, soft shadows or skin tones — you have pushed too far. If you cannot spot the change, you can push further. Looking at the two images side by side at 30% zoom is useless: no artefact is visible at that size.

A recipe that works almost every time

1. Start from the original, not from an already compressed copy.
2. Resize the longest side to the size you will actually use (1920 px for most things; 1280 px for articles).
3. Pick WebP if it is for a website, JPG if it is to send.
4. Set quality to 78–82.
5. Check the result by flipping between tabs at 100%.

Steps 2, 3 and 4 happen in one pass in the image compressor, which puts the maximum-size and output-format selectors right next to the quality slider and shows you how much each file saved.