How favcol works

Every few minutes we do a search on Flickr for new photos tagged “favcol”. This is done using the Flickr API - a way for other people to make use of Flickr’s data.

Each photo is then checked to make sure we can use it. If the “favcol” tag was added by the owner of the photo then we assume this gives us permission. If anyone else added the tag we check the photo’s license - if it has a Creative Commons license allowing derivative works then we use the photo. If neither of these are true, we ignore the photo.

We then analyze each photo to work out the average color.

All the photos are then averaged together, but colors can cancel each other out: very red pictures don’t have much blue and very blue pictures don’t have much red, so on average there’s not much blue and not much red, which can mean green is left over.