How to select the best photo from a burst

Burst mode solved the capture problem and created a selection problem. Every jump shot, group photo and candid now exists as 8 to 20 near-identical frames, and the differences that matter, a closed eye, a soft focus, a photobomber, are invisible at thumbnail size.

So the honest manual workflow is: open frame 1, squint, open frame 2, go back to frame 1, repeat 14 times. Multiply by 40 bursts from one weekend and selection becomes the chore that never happens. Ranking the frames by measurable quality first removes most of that work.

If the burst sits together in one folder, select it there. If the frames are scattered, exports here, phone backup there, pick any one frame and run a similar-photo search with a tight tolerance. PhotoCore pulls the near-identical shots from anywhere in the index, including other drives.

Indexing happens once, in place and read-only, so nothing moves while you work.

Sort the selection by quality, best first. The score comes from Apple Vision image aesthetics on the Neural Engine and reflects sharpness, exposure and framing. The soft frames and the badly exposed ones drop to the bottom on their own.

To be clear about what happens here: PhotoCore does not declare a winner. It orders the candidates so that your decision starts from the top 2 or 3 frames instead of all 14.

Take the top-ranked frames and flip through them at full size with Quick Look. At this stage you are judging the things a score cannot see: the expression, the closed eye, the hand in front of a face, the moment itself.

Two or three frames at full size is a ten-second decision. That is the whole trade of this workflow: the machine narrows, you choose.

Some bursts have a content requirement, and the people filters check it faster than your eyes. Does the frame have the right number of people? Filter for "exactly 2" to find the shot with just the couple, no waiter in the background. With face recognition on, you can also require a specific person: "@anna" plus "exactly 2".

Face detection and matching run on device, and face recognition stays off until you enable it. Details are on the face recognition feature page.

Mark your winner, then handle the remaining frames yourself: move them to an archive folder or delete them in Finder. PhotoCore never removes files on its own, so the leftover frames go exactly where you decide.

The same tools carry the rest of the cleanup. The photo culling feature page covers quality filtering in depth, the guide on removing blurry photos handles the library-wide pass, and finding duplicate photos deals with the copies. For a full real-world case, see organizing wedding photos.