About this site
How FitMyPic tests, writes and corrects its guides
The rules behind every recommendation on this site: what we measure, which sources we accept, what we do when we are wrong, and who pays for the hosting.
At a glance
- Recommendations on this site are backed by tests we run, and the test scripts are published beside the results.
- Numbers are never estimated. If we cannot measure something or find a primary source, the guide says so.
- Encoder versions and dates are stated, because a measurement without them cannot be reproduced.
- Corrections are made in place and described at the bottom of the affected page.
- Advertising pays for the servers. Advertisers have no say in what the guides recommend.
What this site is
FitMyPic is an independent site built around a small set of image jobs that people do constantly and badly: resizing a photo for a form that demands exact pixels, compressing an image under a size limit, cutting a background out of a product shot, and producing a photo that satisfies an authority's rules. The tools run entirely in the browser. Nothing you open is uploaded, which is a deliberate design choice rather than a marketing line — you can watch the network panel while the tool works.
The guides exist because the tools alone cannot answer the questions people arrive with. "Which quality should I use" and "why does my resized photo look worse than the original" are research questions, and they have measurable answers.
How images are tested
Our test method is built for reproducibility, not for impressive numbers.
- The test images are generated by a script we publish.
bench_kit.pycreates a photographic texture, a UI screenshot, flat artwork, a transparent logo, a cutout product photo and a smooth gradient. Generated inputs mean anyone can rebuild an identical corpus. - Every published figure comes from a measurement run.
bench_kit.pyperforms the encodes, computes PSNR and SSIM against a defined reference, and writes the CSV files that each guide links to. - The reference is stated. When we compare downscaling methods, the reference is an area-average (box) resize, and we say so rather than implying that a filter is "correct" in absolute terms.
- Encoder builds and dates are recorded. Current results: Pillow 11.0.0 with libjpeg-turbo and libwebp, measured on 22 October 2026. Different builds will move the numbers slightly, so the scripts are the contract.
- Tool behaviour is verified in a real browser. Anything that depends on your browser — canvas resampling, alpha handling, WebP encode support, memory behaviour with large images — is checked in browsers rather than assumed from documentation.
- Negative results get published too. When a metric stopped improving as file size tripled, that went into the article. A benchmark that only reports the flattering rows is advertising.
Which sources we accept
Format behaviour comes from primary documentation: the JPEG, PNG and WebP specifications and their vendor documentation, browser support data from MDN and caniuse, and library documentation from the projects we actually run.
Requirements that affect real people — passport photo sizes, upload limits, platform re-compression — come from the authority that sets them. For a government photo specification that means the government's own page; for a platform image size it means that platform's help centre. Every such claim carries the URL in the guide so you can check whether the rule changed after we published.
Where no primary source exists, we say that as plainly as we can. "Nobody publishes an official number for this" is more useful to a reader than a confident guess.
Dates, revisions and mistakes
Each guide shows when it was published and when it was last reviewed. A review means the page was reopened and everything that could have moved was re-checked — a platform's pixel requirements, a browser's encoder support, our own measurements. Pages that have not been re-checked recently are labelled with their real review date rather than quietly re-stamped.
If a page contains an error, we fix the error and record what changed. We do not delete inconvenient comments from a measurement and we do not pretend a revision did not happen.
If you disagree with a finding, the most useful thing you can send is a reproduction: the image, the tool or library version, the exact settings, and the result you got. Numbers that fail to reproduce are usually a version difference, and those are worth catching.
What we do not do
- We do not run your images through a server. Every tool on this site processes files in the browser tab.
- We do not recommend a setting we have not tested on the kind of image it is meant for.
- We do not pay for placement in "best tools" lists, and we do not trade links.
- We do not collect accounts, images or personal data to use the tools.
How the site is paid for
FitMyPic is funded by display advertising through Google AdSense, which is what keeps the tools free and account-free. Ads are labelled, are never presented as our recommendations, and do not influence what the guides say — no advertiser sees a draft, and none can ask for a mention. Sponsored material, if it ever appears, will be labelled at the top of the page.
The scripts that licence-plate the ads are the only third-party code on the site. What they collect, and what they do not, is listed in the privacy policy. Your images are not part of that data: they never leave your device.
Contact
Corrections, reproduction attempts, and questions about a specific measurement go to the address on the contact page. The site is edited under the name FitMyPic Editorial; requests to re-run a test with different settings are genuinely welcome, and if your numbers beat ours we will say so and update the page.
Resize, compress and convert images in your browser
Everything happens on your device — no upload, no account.
Sources
- Google Search Central: creating helpful, reliable, people-first content — https://developers.google.com/search/docs/fundamentals/creating-helpful-content
- Google Search Central: quality rater guidelines (PDF) — https://guidelines.raterhub.com/searchqualityevaluatorguidelines.pdf
- libjpeg-turbo documentation — https://libjpeg-turbo.org/Documentation/Documentation
- Pillow image file format documentation — https://pillow.readthedocs.io/en/stable/handbook/image-file-formats.html