Training on an image set of more than one subject using the new method

[SAC020]

I have an instance image set containing a number of persons in a specific position / stance, and I am trying to train for the position, not the specific persons / faces. Is it mandatory to caption with a unique token for each person (face) while using the new method?

Or, as long as I don't plan to "call" any specific subject, but rather the position, I can use a generic unique instance token for all subjects?

I am referring to the statement: The most important step is to rename the instance pictures of each subject to a unique unknown identifier

[TheLastBen]

if you're not training on specific faces, use the same identifier for all images