Credits and licences
Some of what is on this site was made by other people: one photograph, the face detection model, and the typefaces. Here is where each of them came from.
The sample photograph
The picture behind the sample button on the face blurring page is a photograph of three friends walking beside a railway line.
- Photographer
- Priscilla Du Preez, March 2017
- Source
- Published on Unsplash, archived on Wikimedia Commons
- Licence
- CC0 1.0, a dedication to the public domain. No attribution required.
- What we changed
- Cropped to the three of them, resized to 1600 pixels wide, saved as JPEG. The metadata came off in the process, the same way it does for anything this tool exports.
CC0 asks for nothing back. We are crediting her anyway. She gave away a good photograph and it saved us making one.
What CC0 does not cover
CC0 waives copyright. The photographer held the copyright, so she is the one who could waive it, and she did. It says nothing about the three people in the frame.
Their rights are separate, they belong to them rather than to the photographer, and no licence on a stock photography site can sign those away on their behalf. It is the same gap the guide to GDPR and photographs is about, and it applies to us as much as to anyone. Being allowed to publish a picture is not the same as the people in it agreeing to be there.
So we keep the use narrow. The photograph is here to demonstrate a redaction tool and nothing else. It is not an endorsement of this site and it is not in an advertisement. Every copy shown on a page has the faces covered. The uncovered version is only downloaded if you press the sample button, because a tool that covers faces needs a face to cover.
If you are one of the three, or you act for one of them, write to hello@blurmyphoto.com and we will take it down. You do not need to give a reason and we will not argue about what the licence technically allows.
How the covered version was made
Every covered face on this site was covered by the software the site ships. The boxes came from the same BlazeFace detector your browser downloads. The bars were painted by /assets/regions.js, the same file and the same settings as the copy running in your tab.
That includes the miss. The detector found two of the three faces and skipped the one in the middle, whose head is tilted back, so the third box was drawn by hand. We could have used an easier photograph and claimed three out of three. The page on face blurring spends several paragraphs saying the detector will miss faces and that you should check the picture rather than the count, and it would be a strange thing to illustrate with a picture that quietly said otherwise.
The face detector
- Model
- BlazeFace, short-range variant, from Google's MediaPipe
- Licence
- Apache 2.0
- Where it runs
- On your device. It is downloaded when you open a photograph and never before, and your photograph is not sent anywhere for it to look at. How it works sets out how to check that yourself.
Typefaces
Archivo, Instrument Sans and Space Mono, all under the SIL Open Font License 1.1. They are served from this domain rather than from a font CDN, so loading a page here does not tell a third party that you visited.
Everything else
The diagrams, the words, the code and the mark at the top of the page are ours. The terms cover quoting and reuse. For anything they do not cover, send an email.