Mesh rendering using depth estimation

[Waidhoferj]

With depth estimation and a panoramic image, we can render 3D objects that the user can walk around. With panoramic video, we can texture the objects as well. This would make the experience more like an interactive video game instead of a series of panoramas. Here is Ventura's code to do this for cylindrical panoramas, and a SLAM algorithm for tracking camera position.

Look into the techniques and build a toy example on a new branch.