Efficient Geometry-Aware 3D Generative Adversarial Networks - GAN Paper Explained

This course explains the Efficient Geometry-aware 3D Generative Adversarial Networks paper. It covers NeRFs, explicit voxel grids, tri-plane scene representations, pose conditioning, dual discrimination, super-resolution, and multi-view-consistent 3D image generation.

Syllabus

Intro
Tri-plane 3D scene representation intro
NeRF in depth
Explicit voxel grid methods
Tri-plane 3D scene representation explained
Tri-plane is as expressive
Efficient 3D GANs pipeline
Pose correlated facial features
Dual discrimination, super-resolution
Results
Ethical considerations, are we going to change anything?
Intrinsics, extrinsics robustness?
Outro