4RC: 4D Reconstruction via Conditional Querying Anytime and Anywhere
4RC is a unified feed-forward framework for 4D reconstruction from monocular videos with conditional querying across time and views.
project page / arXiv / code / bibtex
Yushi Lan (蘭宇時)
I am a postdoctoral researcher at the Visual Geometry Group, University of Oxford, working with Prof. Andrea Vedaldi.
I received my Ph.D. from Nanyang Technological University, advised by Prof. Chen Change Loy and Asst. Prof. Xingang Pan at MMLab@NTU. Earlier, I earned my B.Eng. in Software Engineering from Yepeida Honors College, BUPT.
About Me
I am interested in the intersection of computer vision, computer graphics, and machine learning. My recent interests include Spatial AI, Embodied AI, Physical AI, 3D AIGC, and world models. I have also worked on shape analysis, inverse graphics, neural rendering, and 3D avatars.
In 2025, I was selected as an Outstanding Prize recipient of the Meshy Fellowship 2025. My Ph.D. thesis, Expanding the Generality of Neural Fields [PDF] , is available online.
News
Recent updates.
- [2026-01] One paper STream3R accepted to ICLR 2026.
- [2025-12] One paper LN3Diff++ accepted to TPAMI.
- [2025-09] One paper WorldMem accepted to NeurIPS 2025.
- [2025-06] Joined the Visual Geometry Group, University of Oxford, as a postdoctoral researcher.
- [2025-02] I received the Meshy Fellowship 2025 Outstanding Prize.
- [2025-02] Defended my Ph.D. at Nanyang Technological University, and my thesis, Expanding the Generality of Neural Fields [PDF] , is available online.
- [2025-02] Two papers SAR3D and 3DEnhancer accepted to CVPR 2025.
- [2025-01] One paper GaussianAnything accepted to ICLR 2025.
- [2024-07] Two papers LN3Diff and Gaussian3Diff accepted to ECCV 2024.
- [2023-07] One paper DeformToon3D accepted to ICCV 2023.
- [2023-06] Joined Google Labs as a research intern.
- [2023-02] One paper E3DGE accepted to CVPR 2023.
- [2023-01] One paper EVA3D accepted to ICLR 2023 as a Spotlight.
- [2022-12] One paper DDF accepted to IJCV.
- [2020-09] Joined NTU MMLab as a Ph.D. student.
- [2020-07] One paper Magnifier accepted to BMVC 2020.
Selected Publications
Selected work from 2020 to 2026.
All Publications | World Models | Spatial Intelligence | 3D/4D Reconstruction | 3D Generation | 3D Avatars | 3D Correspondence | Preprints
Preprint Papers
VFMF: World Modeling by Forecasting Vision Foundation Model Features
VFMF performs stochastic forecasting of future worlds in the latent space of vision foundation models.
project page / arXiv / code / bibtex
Video4Spatial: Towards Visuospatial Intelligence with Context-Guided Video Generation
Video4Spatial studies complex visuospatial reasoning directly from video generation.
project page / arXiv / bibtex
Conference and Journal Papers
STream3R: Scalable Sequential 3D Reconstruction with Causal Transformer
STream3R resolves streaming 3D and 4D reconstruction as a sequential registration task with causal attention.
project page / arXiv / code / bibtex
GaussianAnything: Interactive Point Cloud Latent Diffusion for 3D Generation
GaussianAnything generates high-quality editable surfel Gaussians from images or text.
project page / arXiv / code / demo / bibtex
ArtiLatent: Realistic Articulated 3D Object Generation via Structured Latents
ArtiLatent generates articulated 3D objects.
project page / arXiv / code / bibtex
WorldMem: Long-term Consistent World Simulation with Memory
WorldMem enables long-term memory for video world models.
project page / arXiv / code / demo / bibtex
ObjCtrl-2.5D: Training-free Object Control with Camera Poses
ObjCtrl-2.5D enables training-free object control in image-to-video generation using depth and camera trajectories.
project page / arXiv / code / demo
Textured 3D Regenerative Morphing with 3D Diffusion Prior
A 3D diffusion prior enables textured regenerative morphing for 3D shapes.
project page / arXiv / code / bibtex
SAR3D: Autoregressive 3D Object Generation and Understanding via Multi-scale 3D VQVAE
SAR3D generates and understands 3D objects in a scale-level autoregressive way.
project page / pdf / code / video / bibtex
3DEnhancer: Consistent Multi-View Diffusion for 3D Enhancement
3DEnhancer enhances low-quality 3D assets through multi-view diffusion priors.
project page / arXiv / code / video / bibtex
3DTopia-XL: Scaling High-quality 3D Asset Generation via Primitive Diffusion
3DTopia-XL scales high-quality 3D asset generation via primitive diffusion.
project page / arXiv / code
LN3Diff: Scalable Latent Neural Fields Diffusion for Speedy 3D Generation
LN3Diff creates high-quality 3D meshes from image or text within seconds.
project page / TPAMI / arXiv / code / bibtex
Gaussian3Diff: 3D Gaussian Diffusion for 3D Full Head Synthesis and Editing
Gaussian3Diff uses UV-space 3D Gaussians for view synthesis, animation, and unconditional 3D diffusion for full heads.
project page / arXiv / bibtex
DeformToon3D: Deformable 3D Toonification from Neural Radiance Fields
DeformToon3D achieves high-quality geometry and texture stylization under given styles.
project page / arXiv / code / bibtex
E3DGE: Self-Supervised Geometry-Aware Encoder for Style-Based 3D GAN Inversion
E3DGE is an encoder-based 3D GAN inversion framework for high-quality shape and texture reconstruction.
EVA3D: Compositional 3D Human Generation from 2D Image Collections
EVA3D is a high-quality unconditional 3D human generative model that only requires 2D image collections for training.
project page / arXiv / video / code / bibtex
DDF: Correspondence Distillation from NeRF-Based GAN
DDF studies dense correspondence in NeRF-based GANs and distills it without supervision.
project page / arXiv / Springer / bibtex
Magnifier: Towards Semantic Adversary and Fusion for Person Re-identification
MagnifierNet is a triple-branch network for mining details from whole to parts in person re-identification.
Academic Services
Conference Reviewer: CVPR'21-26, ICCV'23/25, ECCV'24/26, NeurIPS'22-25, ICLR'24-26, SIGGRAPH'23/24/25, SIGGRAPH Asia'23/24/25
Journal Reviewer: TPAMI, IJCV, TVCG, CVIU, TOG
Invited Talks
2025 @ GAMES Webinar: 三维基础模型之重建与生成 [Video]