Introduction
GenRecon3D consolidates emerging directions at the intersection of generative modeling and geometric reconstruction. The workshop targets faithful recovery of underlying 3D and potentially 4D geometry from incomplete observations. Unlike approaches prioritizing visual plausibility, we emphasize metric accuracy, physical correctness, and principled uncertainty reasoning.
The program features invited talks, oral and poster presentations, and community discussions on evaluation protocols and robust benchmarks that assess geometric and semantic accuracy beyond visible surfaces.
Keynote speakers
Gordon Wetzstein is an Associate Professor of Electrical Engineering and, by courtesy, of Computer Science at Stanford University. He is the director of the Stanford Computational Imaging Lab and a faculty director of the Stanford Center for Image Systems Engineering. At the intersection of computer graphics and vision, artificial intelligence, computational optics, and applied vision science, Prof. Wetzstein's research has a wide range of applications in next-generation imaging, wearable computing, and neural rendering systems.
Christian Rupprecht is an Associate Professor of Computer Science at the University of Oxford and a Tutorial Fellow at Magdalen College. His research focuses on unsupervised scene understanding in 2D, 3D, and 4D from images and videos and on representation learning for reconstruction and visual understanding without manual annotations.
Katja Schwarz is an AI Research Scientist at SpAItial working on generative modeling and 3D vision. Her work spans neural representations for 3D inference from sparse observations and generative modeling in both 2D and 3D domains. She previously held research positions at Meta AI and completed her PhD in the Autonomous Vision Group at the University of Tübingen and the Max Planck Institute for Intelligent Systems.
Philipp Henzler is a Research Scientist at Google working on generative 3D AI and controllable video models. He completed his PhD at University College London, where his thesis received the Eurographics PhD Thesis Award. His research includes multi-modal generative modeling and methods related to 3D reconstruction and scene synthesis.
Schedule
Half-day workshop schedule (local CVPR time). Room and streaming link will be added when available.
| Time | Session | Duration |
|---|---|---|
| 13:15 to 13:30 | Welcome and introduction | 15 min |
| 13:30 to 14:15 | Keynote 1: Gordon Wetzstein | 45 min |
| 14:15 to 15:00 | Keynote 2: Christian Rupprecht | 45 min |
| 15:00 to 15:35 | Oral session | 35 min |
| 15:35 to 16:20 | Poster session and coffee break | 45 min |
| 16:20 to 17:05 | Keynote 3: Katja Schwarz | 45 min |
| 17:05 to 17:50 | Keynote 4: Philipp Henzler | 45 min |
| 17:50 to 18:00 | Closing remarks | 10 min |
Paper track
We accept (i) novel full 8-page papers (CVPR 2026 format) for publication in the proceedings, and (ii) shorter 4-page extended abstracts. Extended abstracts may describe novel or previously published work, will not appear in the proceedings, and will be presented during the poster session if accepted.
Important dates
| Milestone | Date |
|---|---|
| Submission opens | February 1, 2026 |
| Submission deadline | March 20, 2026 |
| Notification to authors | March 31, 2026 |
| Camera-ready deadline | April 11, 2026 |
Submission portal: https://cmt3.research.microsoft.com/GENRECON2026/
Organizers
Example papers within scope
The following papers illustrate representative directions aligned with the workshop theme of faithful generative 3D reconstruction. This list is non-exhaustive and provided for reference only.
-
Object-X: Learning to Reconstruct Multi-Modal 3D Object Representations
Di Lorenzo, Gaia, Federico Tombari, Marc Pollefeys, and Daniel Barath. NeurIPS 2025. -
Gen3c: 3d-informed world-consistent video generation with precise camera control
Ren, X., Shen, T., Huang, J., Ling, H., Lu, Y., Nimier-David, M., Müller, T., Keller, A., Fidler, S. and Gao, J., CVPR 2025. -
Flowr: Flowing from sparse to dense 3d reconstructions
Fischer, T., Bulò, S.R., Yang, Y.H., Keetha, N., Porzi, L., Müller, N., Schwarz, K., Luiten, J., Pollefeys, M. and Kontschieder, P., 2025. CVPR 2025. -
Difix3d+: Improving 3d reconstructions with single-step diffusion models.
Wu, Jay Zhangjie, Yuxuan Zhang, Haithem Turki, Xuanchi Ren, Jun Gao, Mike Zheng Shou, Sanja Fidler, Zan Gojcic, and Huan Ling. CVPR 2025.
Contact
Daniel Barath (dbarath@ethz.ch) or Fabian Manhardt (fabianmanhardt@google.com).