Compress images

Drop in your images, choose a quality, and each row shows the size your browser actually produced. Nothing is uploaded.

?
?

Compression runs inside this page using your browser's own image encoder, so your photos stay on your device.

How to use it#

  1. Drop your images onto the box above, or click it to pick them.
  2. Pick an output format, or leave it matching whatever you dropped in.
  3. Move the quality slider until the sizes in the rows look right.
  4. Download the files you want, or take the whole batch.

The size you see is the size you get#

Every number in a row comes from a file your browser has already written. There is no prediction from the dimensions and no rule of thumb, because how well an image compresses depends on what is in it. A photograph of a blank wall and a photograph of a forest at the same resolution can differ by a factor of five.

That matters because the decision you are making is almost always about a limit: an upload form that stops at 2 MB, a mail server that bounces at 10, a page you want to load quickly. Moving the slider changes the numbers in front of you, so you can stop at the first setting that clears the limit instead of guessing and re-checking.

Which format to compress into#

Leaving the format matching the input is the safe choice: a JPG stays a JPG and a PNG stays a PNG, so nothing downstream has to change. It is also the weakest option, because the format an image arrived in is often not the best one for it.

WebP is usually the biggest single win. It handles photographs and flat graphics well, keeps transparency, and every current browser displays it. JPG is the right answer when the file is going somewhere old or fussy. PNG is only worth choosing when you need the image to be pixel-exact, and it ignores the quality slider entirely.

Where the quality goes#

Lossy compression works by discarding detail the eye is bad at noticing: fine gradients, subtle colour shifts, the exact edges of things. At high settings that is genuinely invisible. As you push down, smooth areas like skies and skin start to band, and hard edges pick up a faint halo.

Screenshots and images with small text fall apart much sooner than photographs, because text is nothing but hard edges. If a row looks wrong at 60%, it is usually worth trying WebP at 80% rather than JPG at 80%.

Compressing something twice#

A JPG downloaded from a website has already been compressed, often more than once. Running it through again removes a little more detail and cannot bring back what the earlier pass took, so the result is always slightly worse than compressing the original would have been.

If you still have the file the camera or the design tool produced, start from that. If you do not, a single careful pass here is still far better than repeatedly saving at a low setting.

Frequently asked questions#

Are my images uploaded anywhere?

#

No. Your browser decodes and re-encodes each file on your own device, so the images never travel. The page keeps working with the network switched off, which is the plainest proof of it.

What does Match the input do?

#

It writes each file back out in the format it arrived in, so a JPG stays a JPG. Browsers can read AVIF but cannot write it, so an AVIF is saved as WebP instead, which keeps both the small file size and any transparency.

What quality should I use?

#

Around 80% suits most photographs and is the default. Go higher for screenshots and anything containing small text, where compression shows up around letter edges first. Below about 60% the loss becomes visible on almost any image. PNG is the exception: it is lossless and has no quality to trade, so the slider greys out when PNG is the output, and it skips any PNG that passes through under Match the input.

Why did one of my files get bigger?

#

Some images are already compressed harder than the settings you chose, and a few, such as flat graphics forced into JPG, simply do not suit the format. The row says so rather than hiding it, and the original is still on your machine untouched.

Does compressing remove EXIF data?

#

Yes. Re-encoding through the canvas discards metadata, so the file you download carries no camera model, timestamp or GPS coordinates. That is useful before posting a photo publicly, and worth knowing if you needed that data kept.

Can I make the image smaller in pixels too?

#

Not here. This tool changes how hard the image is compressed and leaves the dimensions alone. The image resizer handles pixel dimensions, including fitting an image under a specific file size.