Lossy vs Lossless

Which format is which

Lossless promises only that the file reconstructs exactly what went into it. It says nothing about what was thrown away before that, which is why a lossless copy of a lossy file recovers nothing.

What lossy actually discards

Not random pixels. A lossy encoder spends its bytes on what eyes are good at noticing and starves what they are not: it keeps brightness detail far more carefully than colour detail, and it smooths fine variation inside areas that are already busy. On a photograph this is close to free, which is why a JPEG at a sensible setting can be a tenth of the PNG and look the same.

On the wrong image it is obvious. Hard edges are the case the encoder handles worst, so text and logos come back with faint halos and speckled noise around every letter. That is not a bad encoder. It is the trade working exactly as designed, on an input it was not designed for.

The expensive mistake: PNG for a photograph

A photograph saved as PNG is often five to ten times the size of the same picture as a JPEG, and it does not look better, because nothing was lost in the JPEG that a viewer can see. The file is larger for no return at all.

The reverse mistake is quieter and worse: a screenshot of text saved as JPEG. It is smaller than the PNG would have been, and the text now carries a grey fuzz that no amount of re-saving will take off.

Re-saving a lossy file costs every time

Every save of a lossy file is a fresh encode, and each one throws away a little more. Open a JPEG, crop it, save, repeat. The damage is invisible on the second pass and unmistakable by the tenth.

The rule that follows is to keep one original and export from it, rather than editing your exports. If a lossy file has to be worked on repeatedly, work in a lossless format and go lossy once, at the end.

Quality settings are not percentages

A JPEG quality of 80 does not mean 80% of the original. It is a dial on the encoder, the scale is not linear, and the same number means different things in different software. What it does do is behave predictably in one direction: visible loss climbs slowly for a long while and then climbs fast.

So the useful method is not picking a number. It is watching the picture and the file size together and stopping just before the difference becomes visible. That point is a good deal lower than most people guess.

Doing it here

The compressor takes a target size or a quality and prints what the file weighed before and after. It runs on your own machine, so nothing is uploaded, there is no account, and there is no cap on how many you put through it.

More guides

• How to Reduce Image File SizeTwo different operations get called this. Most of the time you want both, in a particular order.

FAQ

What is the difference between lossy and lossless compression?

Lossy discards information permanently in exchange for a much smaller file. Lossless keeps everything and reconstructs the original exactly, at a more modest saving. Photographs are the case for lossy, hard edged graphics the case for lossless.

Is PNG lossless?

Yes, always. Saving a PNG repeatedly never degrades it. That is also why a PNG of a photograph is so large.

Is JPEG lossy?

Yes, always. There is no lossless JPEG in ordinary use, and every save is a new encode that throws a little more away.

Does compressing an image make it smaller on screen?

No, not unless you ask for that separately. Compression reduces bytes at the same pixel size. Resizing reduces the pixel size. They are different operations, and it is worth being clear about which one you actually want.

Can I get quality back by converting a JPEG to PNG?

No. The PNG will be larger and will faithfully preserve exactly the degraded picture it was handed. Lossless applies only from that point forward.

What quality setting should I use for JPEG?

There is no universal number, because the scale is not linear and the right point depends on the picture. Watch the image and the size together and stop just before the difference shows.