Is Seedance the only model that supports true reference images?
For example, I like using reference images to insert characters into different scenes. Every other video model calls the reference image "frame 1" but I don't want the picture to be a frame in the video, I just want to take the character in the photo and put them in the requested scene, without the background of the photo being a part of it. Am I just overthinking this or do the others really only build on the environment that is in the photo?