𓇣

Week 08: 3D AI — Neural Radiance Fields (NeRF), Gaussian Splatting, and Mesh Synthesis

Properties2
descriptionSpatial AI systems, 3D Gaussian Splatting, NeRFs, Tripo3D, and generating volumetric meshes from text/image prompts.
tagsdda500, lecture, 3d-ai, gaussian-splatting, nerf, spatial-computing

Description

Transitioning from 2D pixel generation to 3D spatial representations using machine learning. We explore Neural Radiance Fields, 3D Gaussian Splatting, and text-to-mesh reconstruction.

Key Topics

  • NeRFs vs. 3D Gaussian Splatting: Volumetric ray marching vs. real-time differentiable rasterization of 3D ellipsoids.
  • Photogrammetry & Reality Scanning: Mobile scanning (RealityScan) integrated into Unreal Engine and WebGL.
  • Text-to-3D Pipelines: InstantMesh, Tripo3D, and SDS (Score Distillation Sampling).

Graph View

Backlinks