How to compress an image without losing quality

Lossless and lossy, plainly

Lossless means the decompressed image is bit-for-bit identical to the original. PNG is lossless. So is WebP in its lossless mode. The saving comes from describing the same pixels more efficiently, which is why it is modest: there is only so much redundancy to find.

Lossy means some information is discarded, chosen so that what goes is what your eye is worst at noticing - fine colour variation rather than brightness edges. JPEG and ordinary WebP are lossy. This is where the large savings live, and at high quality settings the discarded information is genuinely invisible.

What to do, by what you have

The right answer depends on what is in the picture, not on a general preference.

1. 1Photographs: JPEG at quality 80-90, or WebP. Expect 60-80% smaller with nothing visible lost.
2. 2Screenshots, diagrams, text, flat colour: keep PNG, and reduce the colour count instead. Lossy compression smears exactly the sharp edges these are made of.
3. 3Anything with transparency: PNG or WebP. JPEG has no transparency and will fill it with white.
4. 4Already a JPEG you must not degrade: convert to WebP, which usually saves 25-35% at matched visual quality.

Why re-saving a JPEG repeatedly makes it worse

Each time a JPEG is decoded, edited and saved again, it is compressed again - and the second compression is applied to the artefacts left by the first. The damage accumulates. This is generation loss, and it is why a picture that has been through a chat app three times looks soft in a way the original does not.

Two practical consequences. Keep an original you never re-save, and work from it. And when a tool can copy JPEG data without re-encoding - putting a photograph into a PDF, for instance - that path costs nothing at all.

Common mistakes

Choosing quality 100. It is not lossless; it is lossy with almost no saving. Quality 90 is visually identical and far smaller.

Compressing a PNG screenshot to JPEG. It will be smaller and it will look worse, because text edges are the thing JPEG handles least well.

Compressing twice. Going from the original to your target once is better than stepping down repeatedly.

Frequently asked questions

Is there a truly lossless way to make a photo smaller?

For a JPEG photograph, not meaningfully - it is already lossily compressed, and lossless re-packing saves a few percent at best. For a PNG, yes: reducing the stored colour count is lossless in the sense that the file still reproduces exactly the pixels it now contains, and it can save a great deal on screenshots and flat artwork.

Is WebP better than JPEG?

Usually, for the same visual quality, by roughly 25-35% in file size - and it supports transparency, which JPEG does not. Every current browser reads it. JPEG remains the safer choice when a file has to be accepted by something old or by a form that names it.

What quality setting should I use?

80-90 for photographs is the range where almost nobody can see the difference from the original at normal viewing size. Below 70 it becomes noticeable in skies, skin and smooth gradients. Rather than picking a number, you can give a target file size and let the tool find the highest quality that fits it.