AI Scene Layering

[CoreyArt26165972jfnk]

One idea I had to leverage the power of AI that would be game changing and useful ,would be the ability to take A photo and automatically composite all the elements into layers, even to go as far as filling in the areas you dont see.

So for example imagine a picture of a woman walking in front of a city with cars and buildings. AI could seperate the woman into the top layer, and the city/cars/ect into another layer and fill in the area where the woman was so if you move the woman layer there are no building gaps anymore. Maybe you could select how many composite layers you want to generate or maybe hand select a single element you wish seperate and composite for you to fill in the layers below.

This is also useful for editing elements easier and work more freely.
AI Video already sort of does this, and you could do this manually which is time consuming but instead we are just taking the same technology and using AIs'  compositing abilities in a useful way directly inside a photo editior like photoshop. 

[creative explorer]

@CoreyArt26165972jfnk awesome idea! My students would love the instantaneous nature of AI, but as an educator, I am teaching them the craft—the underlying structure that makes a file usable, editable, and scalable. Your suggestion to have the AI not just generate the image, but also deconstruct it into functional, editable layers, truly would be a game-changer for design education and professional workflow. It would allow students to get the immediate gratification of the AI result while still meeting the core professional requirement of having a clean, layered file. It would be a fantastic bridge, letting designers leverage the speed of AI without sacrificing the quality and editability that Photoshop mastery demands. You are absolutely correct—that feature would save countless hours and prevent a lot of student (and professional) headaches!

[coreyarte]

Thanks! Yeah would be a game changer . Funny how AI videos does it pretty quickly, so dont see why photoshop couldnt do it for a still image. 
As a working professional this would help me for sure. This guy below seems to think so too.
https://www.youtube.com/watch?v=2CTeYAxIY04

[Conrad_C]

It’s a good idea. However if this is something you would need to do soon, it’s already easy. In the demo below, in Photoshop 26.4 (not the public beta) these things are done in a few seconds:

1. Choose the command Layer > Mask All Objects.

2. Copy the original layer into the empty masked layer group that was created for the object you want to isolate. 

3. On the original layer, reselect the person and have Generative Fill remove them. 

Now the person can be moved independently. 

(To do this right, the person’s original shadows, which are still there, should also be removed. That would take just another 2 seconds with the Remove tool.)

What’s good about your idea is automating this some more to make it even easier. However, notice that it takes a while for the generative AI to figure it out. If you want Photoshop to do this automatically for all subjects it thinks it detects in an image (whether you need them or not) could take a very long time, especially on older computers. Think about how long it took for generative AI to fill in the hole behind just the one subject in this photo…and I sped up the demo by 2x.

In some cases, even if your feature idea existed, it might be faster for you to just pick out the few specific subjects you need in a picture using the Object Selection tool, to cut down the number of background holes that you would have to wait for it to fill in. And that’s something you can do right now in the current version.