Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2504.06982 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916681425616896 |
|---|---|
| author | Yang, Yuhang Liu, Fengqi Lu, Yixing Zhao, Qin Wu, Pingyu Zhai, Wei Yi, Ran Cao, Yang Ma, Lizhuang Zha, Zheng-Jun Dong, Junting |
| author_facet | Yang, Yuhang Liu, Fengqi Lu, Yixing Zhao, Qin Wu, Pingyu Zhai, Wei Yi, Ran Cao, Yang Ma, Lizhuang Zha, Zheng-Jun Dong, Junting |
| contents | 3D human digitization has long been a highly pursued yet challenging task. Existing methods aim to generate high-quality 3D digital humans from single or multiple views, but remain primarily constrained by current paradigms and the scarcity of 3D human assets. Specifically, recent approaches fall into several paradigms: optimization-based and feed-forward (both single-view regression and multi-view generation with reconstruction). However, they are limited by slow speed, low quality, cascade reasoning, and ambiguity in mapping low-dimensional planes to high-dimensional space due to occlusion and invisibility, respectively. Furthermore, existing 3D human assets remain small-scale, insufficient for large-scale training. To address these challenges, we propose a latent space generation paradigm for 3D human digitization, which involves compressing multi-view images into Gaussians via a UV-structured VAE, along with DiT-based conditional generation, we transform the ill-posed low-to-high-dimensional mapping problem into a learnable distribution shift, which also supports end-to-end inference. In addition, we employ the multi-view optimization approach combined with synthetic data to construct the HGS-1M dataset, which contains $1$ million 3D Gaussian assets to support the large-scale training. Experimental results demonstrate that our paradigm, powered by large-scale training, produces high-quality 3D human Gaussians with intricate textures, facial details, and loose clothing deformation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_06982 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SIGMAN:Scaling 3D Human Gaussian Generation with Millions of Assets Yang, Yuhang Liu, Fengqi Lu, Yixing Zhao, Qin Wu, Pingyu Zhai, Wei Yi, Ran Cao, Yang Ma, Lizhuang Zha, Zheng-Jun Dong, Junting Computer Vision and Pattern Recognition 3D human digitization has long been a highly pursued yet challenging task. Existing methods aim to generate high-quality 3D digital humans from single or multiple views, but remain primarily constrained by current paradigms and the scarcity of 3D human assets. Specifically, recent approaches fall into several paradigms: optimization-based and feed-forward (both single-view regression and multi-view generation with reconstruction). However, they are limited by slow speed, low quality, cascade reasoning, and ambiguity in mapping low-dimensional planes to high-dimensional space due to occlusion and invisibility, respectively. Furthermore, existing 3D human assets remain small-scale, insufficient for large-scale training. To address these challenges, we propose a latent space generation paradigm for 3D human digitization, which involves compressing multi-view images into Gaussians via a UV-structured VAE, along with DiT-based conditional generation, we transform the ill-posed low-to-high-dimensional mapping problem into a learnable distribution shift, which also supports end-to-end inference. In addition, we employ the multi-view optimization approach combined with synthetic data to construct the HGS-1M dataset, which contains $1$ million 3D Gaussian assets to support the large-scale training. Experimental results demonstrate that our paradigm, powered by large-scale training, produces high-quality 3D human Gaussians with intricate textures, facial details, and loose clothing deformation. |
| title | SIGMAN:Scaling 3D Human Gaussian Generation with Millions of Assets |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2504.06982 |