Repository navigation
- Auto-label (Advanced AI). One click colours a whole folder from the
AI's judgement: red for clear rejects, yellow for repeats, green for the
best frame of a burst, blue for good frames that aren't the pick, and
purple for a best frame that still has a flaw. Undo takes back a whole run,
and Clear removes Auto-label's colours again. Colours you set yourself are
never changed without asking. - Shoot types. Tell OpenCull whether a folder is an event, portraits &
family, or sports & dance, and it changes which faults count and how the
best frame of a burst is chosen. - Show why. Turn it on to see, for each photo, why it was picked or
passed over: for example "ahead mostly on expressions", or "underexposed,
1.5 stops darker than the shots around it". - Better best-frame picks (Advanced AI). A new face reading compares the
faces across a burst, so the frame with the best expressions is put first
more often. - Visual grouping up to 20× faster (Advanced AI). With "Group visually
similar photos" on, OpenCull's own small scene network now settles the
easy cases itself and asks MobileCLIP only about the photos it is unsure
of. The groups come out the same. - Faster scans. Portrait photos are stood upright twice as fast, so a
face scan finishes about 7% sooner, with exactly the same results. - A tidier toolbar. An AI button (with Advanced AI on) opens a bar with
the shoot type, Show why, Auto-label and Clear. Speed Cull is a lightning
bolt you can click to turn off, and the "Ready" sign is gone. - Worth knowing: folders already scanned with Advanced AI are scanned
once more the first time you open them, for the new face reading. Basic AI
is unchanged.
See the README for what works, what does not yet, and how to open the app the first time.