Opensolr Photos 3.1: Similar Photos That Really Are Alike

A group of eight thousand photos is not a group of duplicates. This release makes the view say something true, and say it quickly.

01 Groups that stay apart

Groups used to be joined whenever they shared a photo: one group with a second, that one with a third, and on a loose setting the chain ran through the whole library until a single group held nearly every photo you own. Each group is now what the index actually found, and a photo is shown in the first group it belongs to, not in five.

02 A group of more than fifty is a category, not a copy

When a key is held by hundreds of photos it is describing a subject — sky, outdoor, a person — not a set of copies. Those are left out now, so what you scroll through is what you might actually delete.

03 The slider says something useful at both ends

The two loosest stops are the same first 2 and same first 3 words. The old any 2 words and any 3 words stops are gone: on a real library half the photos share two ordinary words, so the stop could only ever answer with one enormous group. From there it tightens to any 4 words, all 5 words, the same photo by EXIF, then file name, file size and the exact same file.

04 One request per screen

The groups arrive in a single request and are held for as long as you chose in Me, so walking the slider back over a setting you have already seen costs nothing. Each page of the grid is one request as well — up to twenty groups and at most three hundred photos — instead of the forty it took to fill a screen when one group could hold thousands.