I tryed so many nodes combinations and tutorial videos on yt, but the result seem to never meet what i need. Here’s the deal: i want to give comfy an image of a person (img1) and say to it “hey, this area, right here (one arm, for instance), i want you to act ONLY on this area and make it look like the style of this other image (img2 - a robot arm, for instance)”. I don’t want the AI doing things to the rest of img1, only the place i indicate. how do i do that with ip adapter?