Why this exists
Every online image converter works the same way: you upload a file, a server processes it, you download the result. That design has consequences you feel every time. You wait twice, for the upload and the download. You run into a file size cap and a daily quota, because the operator pays for every conversion. And your image sits on a stranger’s disk for some length of time you have to take on faith.
None of that is necessary any more. Browsers have been able to run compiled code at close to native speed since WebAssembly shipped in 2017, and the same encoders desktop software uses — mozjpeg, libwebp, libavif, libpng, oxipng — compile to it perfectly well. This site runs them in a background thread on your own machine. There is no upload because there is nowhere to upload to.
The economics follow from the architecture. Processing costs nothing, so there is no reason for a file size limit, a daily quota, an account, a watermark or a paywall on basic conversion. Those limits exist elsewhere because someone is paying for CPU time.
How to check the privacy claim
“We never see your files” is easy to write and impossible to verify on most sites. Here it is enforced by the browser rather than by our good intentions: every page is served with a Content Security Policy that forbids it from opening a network connection to any host but its own. If the code tried to send your image somewhere, the browser would block the request.
Confirming it takes about ten seconds. Open any tool, press F12, switch to the Network tab, and convert a file. You will see the page fetch its own code and the relevant codec, then go completely silent while it works. There is an automated version of this check in the project’s test suite too, run on every change, which fails the build if any request leaves carrying enough data to contain an image.
The rules behind the writing
A site with hundreds of near-identical pages is usually worth nothing to read. These are the rules that are meant to keep this one from becoming that:
- Numbers are measured, not estimated. Every file size comparison comes from running the actual encoders over a fixed test corpus. Tables that have not been measured yet are not displayed at all, rather than filled with plausible guesses.
- Anything that can go stale carries a date. Browser support tables, platform specifications and file size limits all say when they were last verified. If another site does not tell you that, its numbers may be years old.
- Pages say when the tool is the wrong choice. Every conversion page has a section on when not to use it. A converter that only tells you to convert is an advertisement.
- A page ships when it has something specific to say. Several high-traffic conversions are deliberately unpublished because the tool would silently do the wrong thing — dropping pages from a document, or animation from an image. A page that does not answer the question behind the search is worse than no page.
Who runs it
Pixgrain is built and maintained by Alexandr Stoianov, an individual rather than a company, from Moldova. Corrections are genuinely welcome — if a support table is out of date, a measurement looks wrong, or a conversion behaves in a way the page does not describe, get in touch and it will be fixed and the date updated.
The terms of use are short, for the same reason the privacy policy is: a site that runs on your own machine does very little on your behalf, so there is not much to govern.
What it costs
Nothing, and there is no account to create. The site may carry advertising in future to cover hosting; if that happens, the privacy policy will be updated before it goes live rather than after, and consent will be asked properly. The core promise — that files are processed on your device and never transmitted — is architectural and is not up for revision.