Course positioning
CS 348N studies how to generate and manipulate high-quality 3D models and scenes with machine learning. It covers geometry representations, 3D datasets, point-cloud/mesh networks, neural implicits, 3D generative models, conditional generation, and evaluation — a core course for 3D generation/graphics.
What it covers
- Geometry representations: point clouds, meshes, voxels, SDFs, NeRF
- 3D data: ShapeNet, Objaverse, scene datasets
- Point-cloud/mesh networks: PointNet, GCN, MeshCNN
- Neural implicits: SDFs, occupancy, NeRF
- 3D generation: VAE/GAN/diffusion/autoregressive
- Conditional generation: text/image to 3D
- Evaluation and synthetic data
Prerequisites
CS 148; deep learning basics; CS 229 recommended.
Assessment & workload
Assignments plus a project. Check the current syllabus for details.
Who it's for
3D generation, graphics, AR/VR, and robot simulation.
Study advice
Learn NeRF and diffusion first, then revisit 3D versions of GAN/VAE. Pair with CS 468.
Related courses
Official course description
Course Description: Generation of high-quality 3D models and scenes by leveraging machine learning tools and approaches. Survey of geometry representations. Public 3D object and scene data sets. Neural architectures for geometry, including deep architectures for point clouds and meshes. Generative models for 3D: autoencoders, GANs, neural implicits, neural ODEs, autoregressive models. Conditional generation based on images or partial geometry. Variation generation. Evaluation metrics for content generation. Use of synthetic data in ML training pipelines. Prerequisites: CS148 and the rudiments of deep learning. Recommended: CS229.