we don't even need AI for that: https://3dqlab.stanford.edu/what-is-volume-rendering/