Allow Photoshop to manipulate objects in a 3 dimensional space and then convert the image back into a 2 dimensional space using AI and user input. Allow Photoshop AI to generate a 3d model and objects including light sources and distant light points from the image and allow users to turn the image along the z axis so that they can modify/change objects in 3d space and then convert the image back to 2d space or export an image of the 3d space based on user input and AI. This would allow someone to drag an arm up and out or behind their back etc. This would also mean the AI has to be intelligent enough to
move an arm, leg, or head around by a joint and respect 3d object boundaries. This would be an amazing tool that let you reshoot an entire image and do anything your imagination can come up with. The AI should be smart enough to wrap the "back" of a 3d object in a texture based on AI data and information in the original image.