Nan Liu
I work on generative models. Currently at Roblox, previously Morpheus AI (acq. by Roblox), World Labs, and Google.
Products
- 2025Marble: Multimodal World ModelWorld Labs
- 2025World GenerationWorld Labs
- 2024Virtual Try-OnGoogle
Papers
- 2026Back to Pixels: Lifting Latent Video Diffusion into Pixel SpacearXiv
- 2026MPEG-Diffusion: Mixed-Resolution Video GenerationarXiv
- 2024Fashion-VDM: Video Diffusion for Virtual Try-OnSIGGRAPH Asia
- 2024Compositional Image Decomposition with Diffusion ModelsICML
- 2024M&M VTO: Multi-Garment Virtual Try-On★CVPR
- 2023Unsupervised Compositional Concepts DiscoveryICCV
- 2022Composable Diffusion ModelsECCV
- 2021Learning to Compose Visual Relations★NeurIPS