Open shawnlauzon opened 13 years ago
Unfortunately, those favorites groups don't have anything very interesting. Perhaps what we could do is download just the data, including the number of favorites, order them by number of favorites, and then just download pictures page by page. This is slow, but perhaps that's because of all the extra data we download. If we only download the ID and the URL it might be pretty fast.
Groups are such a pain; there are tons of them and they tend to show the same pictures. There's also no easy way to find all the best groups.
Using the favsXX groups might be a good idea http://www.flickr.com/groups/favs10/ and then don't have to order them so much (seems to go up to favs500) -- but then there's "800faves" -- it's a bit ridiculous
Also, possibly the machine tags could help. http://www.flickr.com/groups/api/discuss/72157594497877875/
A huge benefit of tags is that you can search for multiple tags in a single search, whereas for groups you have to do a single group.
However, tags don't seem to describe how good the picture is, so we end up with a ton of CRAP.
There seem to be two pretty decent sets of favorites:
If we just loop through these favorites, we can do it pretty fast; start from the most favorites, then keep querying until we get a decent number of photos. If those are not enough, user can choose to download more pictures which go lower.