workflow – How to separate different subjects images automatically?

This is a generally good question — I’d always be interested in an efficient way to “pre-edit” during the shoot. The bad news is I don’t think there’s any easy solution, at least for canon.

My suggestions:

  1. Change your goal from tagging all pictures in a group, to marking the first picture of a group. This can be done with the camera, or in an old-fashioned way.

Camera: On the camera playback mode, assign 1 star to the first picture taken from each group. This is many clicks, but it is a true metadata change to identify a photo. There may be a way to assign the star rating through a custom function or another set-up, but I didn’t find any obvious methods when looking.

Old-fashioned: Signal a transition in the album by taking a picture of an all-black or all-white something. Better, make a cue card and write the relevant meta-data on it for each object shoot. Take this picture before each new shoot, or after, but be consistent. On post-production you still need to visually group the photos. To me this is the easiest and most reliable solution.

  1. Write a script.
    timestamp evaluation
    Let’s assume there is an added delay for tear-down and set-up between shoots. You could write a script that reads the datetime stamp metadata after import, and identifies gaps greater than X minutes in the time the photo was taken, and bins the photos based on this. One better is a function that estimates boundaries between shoots, by identifying a delay that is 2-4 standard deviations longer than the average delay. Head over to the coding stacks for how to make this more robust or customized to your studio’s workflow.

detect all-black picture
Assume you take an all-white or all-black photo between objects, suggested in part 1. On import, (write and) run a script that filters for photos by exposure threshold–a high one if all-white picture is taken, a low one if all-black. Then, pictures between each identified all-X picture are put into respective folders.

The first scripting solution can likely be done with just about any coding language and no special downloads. This one would require some level of photo processing, but would be more reliable.

  1. AI software on computer.
    Google Photos, Apple, and I think Amazon all have some level of object-recognition software that will sort your pictures. It’s likely part of many professional photo management software too. It won’t be fool-proof, but it would probably get within a 95% accuracy, depending on differences in your objects.

Finally,
Some marginal details that could be relevant for other, slightly different use cases: Some cameras and/or software will put photos into separate folders by date and/or geo-location (must be on camera). Useful only if such details are changing between shoots, but at least captures coarse-grain grouping of photos.