Gen3R: 3D Scene Generation Meets Feed-Forward Reconstruction

Jiaxin Huang1 Yuanbo Yang1 BangBang Yang2 Lin Ma2 Yuewen Ma2 Yiyi Liao1,†
1Zhejiang University 2ByteDance
†Corresponding Author
Paper Code arXiv

TL;DR: Create multi-quantity geometry with RGB from images via a unified latent space that aligns geometry and appearance.

Abstract

We present Gen3R, a method that bridges the strong priors of foundational reconstruction models and video diffusion models for scene-level 3D generation. We repurpose the VGGT reconstruction model to produce geometric latents by training an adapter on its tokens, which are regularized to align with the appearance latents of pre-trained video diffusion models. By jointly generating these disentangled yet aligned latents, Gen3R produces both RGB videos and corresponding 3D geometry, including camera poses, depth maps, and global point clouds. Experiments demonstrate that our approach achieves state-of-the-art results in single- and multi-image conditioned 3D scene generation. Additionally, our method can enhance the robustness of reconstruction by leveraging generative priors, demonstrating the mutual benefit of tightly coupling reconstruction and generative models.

Interactive Examples

Loading interactive demo...

Comparisons

More Results

BibTeX


      @misc{huang2026gen3r3dscenegeneration,
          title={Gen3R: 3D Scene Generation Meets Feed-Forward Reconstruction}, 
          author={Jiaxin Huang and Yuanbo Yang and Bangbang Yang and Lin Ma and Yuewen Ma and Yiyi Liao},
          year={2026},
          eprint={2601.04090},
          archivePrefix={arXiv},
          primaryClass={cs.CV},
          url={https://arxiv.org/abs/2601.04090}, 
      }