I want to get a 3d model (even if it's not perfect) from a few photos that are already taken (for example, from a film). I would get some frames of a place in a film and then I want to get the 3d Model.
I've searched some tools and I've found Smoothie 3D which helps you to create the model from a photo but it would work for a squirrel for example, but if there's a room with two actors inside and they're moving, I would like to get the room in 3d as well as the actors but in separated models so I can animate the actors inside the room.
I have also tried photoscan but it doesn't work too good if the actors are moving because it doesn't understand what model it should get (the room, one shape of the actor, etc).
I was thinking about some script to remove people from all the frames, then get the 3d model of the room and somehow get the 3d models of the actors later (I don't know how).
Does anyone know a better pipeline or a tool that could do the trick for that? or maybe any deep learning algorithm to get the room ignoring the moving actors and the same for the actors?
You can try this pipeline :)