Chris Rockwell

I am an AI Scientist at Roblox working on Roblox Reality, with the goal of enabling photorealistic gaming using multimodal generative AI. Previously, I was part of the founding team at Morpheus AI under Xun Huang, which was acquired by Roblox.

I received a Ph.D. in Computer Science from the University of Michigan and was advised by Justin Johnson and David F. Fouhey. During my Ph.D., I interned at NVIDIA Deep Imagination Research with Chen-Hsuan Lin, and at Meta Reality Labs with Hung-Yu Tseng.

My research works towards simulating 3D-consistent worlds, contributing to large-scale video curation, camera pose estimation, 3D asset generation and neural rendering. My work at Roblox and Morpheus applies this focus to real-time video generation and world models.

GitHub  /  Google Scholar  /  CV  /  LinkedIn  /  X

We introduce DynPose-100K, a large-scale video dataset of dynamic content with camera annotations.

Our flexible method produces accurate and robust pose estimates using complementary strengths of Correspondence + Solver and Learning-Based methods.

Pretrained models are highly effective at 3D captioning; we use them to collect large-scale 3D-text dataset Cap3D and finetune text-to-3D models.

Small modifications to a ViT enable computations similar to the Eight-Point algorithm, and are a good inductive bias for pose estimation.

Our forward warping-based method enables real-time, high-quality novel view synthesis on novel objects from sparse views.

Combining 3D reasoning and autoregressive modeling facilitates high-quality and consistent single-image novel view synthesis.

Template