What approach is the best to take for a cv-beginner trying to learn 3d-reconstruction?

I'm currently in the second year of studying CS in uni, and as part of a project course I have just started reading about computer vision, and I'm planning to do a project involving 3d-reconstruction. From my understanding so far, there exist more classical cv-approaches to this problem, such as Structure from Motion, while more modern approaches using deep learning and in particular radiance fields are showing better and better results. So my question is: what would a wise approach …

beginner computervision cv