Dr. Aleksander Hołyński (Google): “Recent Advances in 3D Generative AI”

Speaker: Dr. Aleksander Hołyński (Google)

Title: How I Learned to Stop Worrying and Love the Data Monster

Video: Click here to view

Abstract:  Recent advances in visual generative models have led to the generation of high quality, diverse images and videos of nearly any imaginable concept, thanks to increasingly large models and huge training datasets. Looking at the quality of images that these models are able to produce, one may wonder: “are we done?”. In this talk, I argue that we’ve only just scratched the surface. These models have seen billions of training images—what other knowledge have they amassed along the way? And what else can they be used for, beyond the generation tasks they were trained for? This talk will cover recent explorations in answering these questions, showing how the outputs of large text-to-image models can replace curated datasets for downstream supervised tasks, like image editing and 3D reconstruction. Delving deeper, we’ll see that by probing the internal representations of these models, one can extract underlying properties of the generated content, such as scene geometry, point correspondence, and more.

Bio: Aleksander Holynski is a senior research scientist at Google DeepMind and a postdoctoral scholar at Berkeley AI Research, working with Alyosha Efros and Angjoo Kanazawa. He received a PhD in Computer Science from the University of Washington, where he was advised by Steve Seitz, Brian Curless, and Rick Szeliski, and a B.S. with High Honors from the University of Illinois at Urbana-Champaign. His research focuses on computer vision, computer graphics, machine learning, and generative models. His work has been recognized with best paper awards at ICCV 2023 and CVPR 2024.