PRACTICAL GUIDE
How to Reduce Image File Size Without Ruining Quality
Why the dimensions matter more than the quality slider, which format suits which picture, and how to tell when you have stopped gaining anything.
Last updated
Start with the dimensions, not the quality slider
Most oversized images are oversized because they are simply too big. A photo straight off a phone is often four thousand pixels wide; the place it is going shows it at twelve hundred. Every pixel beyond what is displayed costs bandwidth and buys nothing.
This is also where the arithmetic is on your side. Halving the width and the height leaves a quarter of the pixels, and the picture looks identical at the size it is actually viewed. No amount of fiddling with a quality slider produces a saving that large without visible damage.
Then pick the format that suits the picture
JPG is built for photographs: gradual changes in tone compress extremely well, and the artefacts it produces hide inside detail. PNG is built for the opposite: flat colour, sharp edges, screenshots, logos, anything with text in it. Using the wrong one costs you either quality or size, sometimes both.
WebP usually lands somewhere between a fifth and a third smaller than a JPG of comparable quality, and it handles transparency, which JPG cannot. Every current browser reads it. The one place to hesitate is a file someone will open in older desktop software.
Compress in steps, and judge at the size people see
Going straight to a low quality setting is how images end up looking cheap. Come down in steps and compare at the size the image will actually be displayed, not at two hundred per cent in an image viewer.
For most photographs the last clearly invisible step is somewhere around eighty to eighty-five per cent quality. Below that you start trading appearance for bytes, and around blocks of flat colour or text the trade shows up first.
When the file simply will not get smaller
A JPG that has already been compressed once has had its redundancy removed. Compressing it again mostly adds a second generation of artefacts and saves very little, which is why our compressor gives you the original back untouched when it cannot beat it rather than handing you a worse file of the same size. When you see that message, the honest next move is to reduce the dimensions instead.
Keep the original
Treat the file that came out of the camera as a master and work on copies. Compression is not reversible: once the detail is gone, no tool restores it, and you will eventually need a larger export, a different crop or another format. Keeping one full-quality source costs disk space you almost certainly have.
Frequently asked questions
What quality setting should I use?
Around eighty to eighty-five per cent is invisible for most photographs and cuts the file substantially. Go lower only for images that will be shown small, and check the result at its display size before deciding. Screenshots and graphics with text are the exception: they show compression damage much earlier, and are usually better served by PNG or WebP.
Why did my PNG get bigger when I compressed it?
PNG compression is lossless, so a tool that re-encodes a PNG has no quality to trade away and can easily produce a slightly larger file than the original encoder did. For a photograph the answer is to convert to JPG or WebP; for a graphic, reducing dimensions is the reliable saving.
Does resizing lose quality?
It discards pixels, which is a real loss, but not one you can see if the result is still at or above the size it will be displayed. The mistake is resizing below that and then having to enlarge again later, which no tool can do without inventing detail.
Should I just convert everything to WebP?
For anything shown in a browser, usually yes. It is smaller than JPG at comparable quality and unlike JPG it supports transparency. Keep a JPG or PNG when the file will be opened in older desktop software, attached to an email, or handed to someone whose tooling you do not control.