机器学习与数据科学博士生系列论坛（第六十二期）—— A Gentle Introduction to Text-to-3D Techniques

摘要：
There has been a surge of demand for detailed and controllable 3D digital assets, driven by the needs of simulation, multimedia experience, and artistic creation. In practice, we expect convenient generation and manipulation of 3D content, particularly in alignment with text prompts. Recently, text-to-3D techniques have made strides and among them, 2D lifting methods inspired by Score Distillation Sampling (SDS) show great promise. 

In this talk, we will introduce the basics of 3D scene representation and delve into how text-to-3D generative models synthesize high-quality 3D content with the strength of the 2D diffusion model. We will then briefly discuss the interaction between 3D representation, generators, and knowledge distillation.

论坛简介：该线上论坛是由张志华教授机器学习实验室组织，每两周主办一次（除了公共假期）。论坛每次邀请一位博士生就某个前沿课题做较为系统深入的介绍，主题包括但不限于机器学习、高维统计学、运筹优化和理论计算机科学。