See every file before and after, cap the dimensions, strip the location data you did not mean to publish. Sizes come from the files the server actually produced.
Supported input formats — pick one to select a file.
Compress an image to a smaller file without changing what it is. Drop up to 100 MB of JPG, PNG, WebP, GIF or TIFF, set a quality level and cap the dimensions, and see every file's size before and after. Each file keeps the format it came in with — moving a file between formats is what the Image Converter is for, and this page deliberately does not duplicate it.
If a compression cannot beat the original, this page hands back the original, shows the original's size, and says why — rather than linking a larger output because a run is supposed to look successful. That is not a rare edge: a JPEG already saved at a low quality has little left to give, and a lossless re-encode of an already-efficient PNG can come out slightly larger. A tool that ignored this would be quietly serving you a worse file than the one you handed it.
The quality level is the encoder's own quality. The dimension cap resizes an image larger than the cap and never scales one up. Metadata is removed by default: GPS coordinates and camera details are stripped unless you switch the keeping option on. That default is deliberate rather than incidental — the usual reason a photo gets compressed here is that it is about to be published somewhere.
The encoder runs on the server, so the image has to be uploaded for the page to work at all. That is worth stating plainly, and it is why this entry's privacy note reads differently from the seven generator tools': there is no version of this tool that keeps the bytes on the device. The converted result is held in ephemeral storage only long enough to be downloaded, and a scheduled cleanup job purges converted files once the retention window passes.
Your image is uploaded to this server and compressed there, because the encoder runs on the server — this page cannot work with the file staying on your device, and it does not claim to. The converted result is kept in ephemeral storage only so that you can download it, and a scheduled cleanup job purges converted files when the retention window passes; that window is 24 hours by default on this deployment, and the paid plans advertise a longer download window. No account is needed to use the tool.
It depends on the file, and the page will not promise you a figure. What a compression can reach is set by how the image was encoded in the first place and by the quality level you choose: a photo as it left a camera has a lot of room, while a JPEG that is already heavily compressed, or a PNG of flat colour, has very little. In one measured run a 4,098,701-byte JPEG stored at quality 100 came back at 253,778 bytes — 93.8% smaller — but that was a single high-entropy test image at quality 30, so it is a worked example and not a prediction for your file.
No — there is no target-size field, and the page will not pretend otherwise. You set a quality level and the output is whatever the encoder produces at that level for that image; the only way to land on a size is to try a level, read the resulting figure in the row, and try another. A target-size mode would have to run the encoder repeatedly until it hit a number, which would make the result differ unpredictably from one image to the next. If what you actually need is a ceiling rather than a compression, capping the dimensions usually gets you closer than lowering quality, because quality has far less effect once an image is already compressed.
For JPEG, GIF and TIFF, yes at any quality level below the maximum — those formats cannot store the image without discarding detail, and how much they discard is exactly what the quality setting decides. For PNG, WebP and AVIF the encoder has a genuinely lossless path, where the image is written out with nothing discarded and the saving comes from writing it more efficiently rather than from throwing anything away. So the honest answer depends on the format of the file you put in, which is one reason the page cannot make a blanket promise about quality.
Because the compression could not beat the original, and the page will not hand you a bigger file to make a run look successful. When that happens the row keeps the original, shows the original's size, and says why. It is common rather than exceptional: an image already saved at a low quality has nothing left to give, and a lossless re-encode of an already-efficient PNG can end up slightly larger. Lowering the quality further will not help in that case — the format or the dimensions are what would.
No, every file keeps the format it arrived in. The output-format picker was removed on purpose, because a separate Image Converter already covers moving a file between JPG, PNG, WebP and the rest — two pages offering the same format switch would compete for the same search and the same visitor. Use this page when the file should stay what it is and simply be smaller.
100 MB per image, and images only, with two files processed at a time. A file over the limit, or one that is not an image at all, is declined with a message naming which of the two it was rather than a generic failure. If you are signed out there is also a daily limit on how many conversions the platform accepts from one visitor.
