I'm looking to composite an opaque object (e.g. 3D art) into video footage of semitransparent fog/clouds before a landscape background so that the object seems immersed in the clouds but occludes the background. Ideally this would mean separating the footage with the clouds and landscape into two layers.
What is the state of the art in matting/masking/extracting transparent objects from video frames? Any depth mask/inter-frame/AI algorithms available? I'm a software developer, but any suggestions for existing (experimental) tools are also welcome.