Skip to main content
Known Participant
February 13, 2026
Open for Voting

P: Improve Select Subject to use the "select people" without detecting a face

  • February 13, 2026
  • 1 reply
  • 41 views

The “select people” is great, it’s amazing, because you can automate the selection of clothes, shoes, accessories and things like that… 

However, it only works if PS finds a face first…

Sometimes, users need to use that tool but the photo is cropped and not showing a face, so it won’t detect any face, therefore not showing those options, “but they’re there”…

People consider allowing us to select those things even if Photoshop didn’t detect a face first

    1 reply

    c.pfaffenbichler
    Community Expert
    Community Expert
    February 20, 2026

    Interesting problem. 

    I can understand the »head« being used as a kind of primary identifier of a person in an image. 

    And I have no idea how difficult it would be to amend this approach, so I am not terribly optimistic on it being remedied in a timely fashion. 

     

    How frequently do you face the issue? 

    As a work-around one could use Generative Fill to have it create a head, which it might be possible to automate somewhat. 

    ril3ydxAuthor
    Known Participant
    February 20, 2026

    And I have no idea how difficult it would be to amend this approach, so I am not terribly optimistic on it being remedied in a timely fashion. 

    Well, it already needs to “find” the clothes, shoes and all, the head is only the first activator, let’s’ say, i think they would just need to add a button or lemme select them without detecting the face, it won’t change much
     

    As a work-around one could use Generative Fill to have it create a head, which it might be possible to automate somewhat. 
     

    That doesnt make a lot of sense

     

    c.pfaffenbichler
    Community Expert
    Community Expert
    February 20, 2026

    First off I forgot to mention that I can reproduce the issue, but that’s no surprise. 

     

    Well, it already needs to “find” the clothes, shoes and all, the head is only the first activator, let’s’ say, i think they would just need to add a button or lemme select them without detecting the face, it won’t change much

    I am not sure how the algorithm actually works – I imagine it might be possible that it determines »head« and expands the »search-area« to determine the identifiable areas from there. 

    Otherwise a T-shirt the floor might also qualify as a person (just missing certain parts).

     

    That doesnt make a lot of sense

    I suppose it depends on how frequently the problem arises and how quick one is in masking the intended areas by hand oneself. 

    And as a work-around it would probably need manual input anyway.