[jrussino]

> I mean understanding the 3d scene in detail

For “classical” (non-ML) computer vision/photogrammetry, you may be interested in this text by Hartley & Zisserman:

https://www.cambridge.org/core/books/multiple-view-geometry-...