You pressed the shutter once and the camera took twelve frames. A similarity scan sees twelve copies of one photo. You see twelve different moments - and exactly one of them has everyone smiling with their eyes open.
Burst mode is where duplicate cleanup goes wrong most often, because burst frames defeat the idea of a duplicate completely. Each frame was exposed separately: the ball a fraction higher, the blink finished, the smile arrived. Burst shots are not duplicates. They are a short film of stills, and their whole value is the tiny differences between them. A hash based exact scan knows this and leaves bursts alone - every frame has a different hash. A similarity scan does not know it, and its bulk delete will cheerfully keep an average frame and erase the one frame you would have framed. This page is the honest case for treating bursts as a set you curate by eye, slowly, or not at all. Safety frame, as always: backup first, review before delete, and stage anything you remove for a week in the Recycle Bin or a staging folder.
Nobody shoots a burst to get twelve copies. You shoot it because the moment is moving - a child blowing out candles, a dog mid jump, a group that will not hold still. The camera trades storage for timing, and the best frame is usually not the first one. It might be the only frame in focus, the only one without a blink, the only one where the ball is in the air.
That is why a percentage score is meaningless here. A burst frame can be 99 percent similar to its neighbour and 100 percent different in the one detail that matters.
When a tool auto-selects within a similar group, it typically keeps the first file, or the largest, or the one in the middle - a filing decision, not a photography decision. It has no idea which frame is sharp at the face, which one caught the expression, which one you would print.
Run that on a wedding burst or a baby's first steps and the loss is invisible until the day you zoom in - weeks later, after the Recycle Bin is empty. This is the exact failure the staging rule exists to catch.
If burst folders are eating real space, thin them like a photographer, not like a cleaner.
The full cleanup, from the first scan to the final delete, is shown live in the video.
Watch Full VideoNo. Burst frames are separate exposures with different file hashes, so an exact match scan will not group them. If they are grouped, your tool is in similarity mode - treat that list as viewing only and choose keepers by eye.
Yes, manually. Pick the best frame or two at full size, secure them in your album, then stage the rest in a folder for a week or more before deleting. The danger is only in bulk or automatic deletion, which cannot judge the frames.
Often less than people fear - a burst of twelve phone photos is commonly under 50 MB. Compare that with a few duplicate videos or one WhatsApp swamp before risking frames you can never reshoot.