Seed3D 2.0: Advancing High-Fidelity Simulation-Ready 3D Content Generation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911683248652288 |
|---|---|
| author | Gu, Diandian Lin, Jing Liu, Gaohong Liu, Jiahang Ma, Su Shi, Guang Wang, Jun Wang, Qinlong Wu, Qianyi Xu, Zhongcong Yi, Xuanyu Yu, Zihao Zhang, Jianfeng Zheng, Zhuolin Zhu, Yifan Chen, Rui Guo, Hengkai Guo, Xiaoyang Han, Mingcong Han, Xu Li, Xiu Liang, Yixun Lou, Weiqiang Lu, Junzhe Luo, Guan Qin, Minghan Wang, Shuguang Wang, Yuang |
| author_facet | Gu, Diandian Lin, Jing Liu, Gaohong Liu, Jiahang Ma, Su Shi, Guang Wang, Jun Wang, Qinlong Wu, Qianyi Xu, Zhongcong Yi, Xuanyu Yu, Zihao Zhang, Jianfeng Zheng, Zhuolin Zhu, Yifan Chen, Rui Guo, Hengkai Guo, Xiaoyang Han, Mingcong Han, Xu Li, Xiu Liang, Yixun Lou, Weiqiang Lu, Junzhe Luo, Guan Qin, Minghan Wang, Shuguang Wang, Yuang |
| contents | We present Seed3D 2.0, an advanced 3D content generation system built on Seed3D 1.0, with substantial improvements across generation fidelity, simulation-ready capabilities, and application coverage. For geometry, a coarse-to-fine two-stage pipeline decouples global structure learning from high-frequency detail recovery, while a locality-aware VAE achieves higher spatial compression and more efficient decoding. For texture and material generation, we replace the cascaded pipeline of Seed3D 1.0 with a unified PBR model that directly generates multi-view albedo and metallic-roughness maps, enhanced by Mixture-of-Experts scaling and VLM-based semantic conditioning for improved material precision and visual fidelity. Beyond single-object generation, Seed3D 2.0 introduces a simulation-ready model suite comprising scene layout planning, part-aware decomposition, and training-free articulation generation, enabling coherent scene construction and part-level physical interaction across physics and graphics engines. A large-scale human preference study against five recent commercial models shows that Seed3D 2.0 achieves consistent win rates of 69.0% to 89.9% in textured 3D asset generation. Seed3D 2.0 is available on https://exp.volcengine.com/ark/vision?_vtm_=0.0.c70961.d701978.0&mode=vision&modelId=doubao-seed3d-2-0-260328&tab=Gen3D |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_13862 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Seed3D 2.0: Advancing High-Fidelity Simulation-Ready 3D Content Generation Gu, Diandian Lin, Jing Liu, Gaohong Liu, Jiahang Ma, Su Shi, Guang Wang, Jun Wang, Qinlong Wu, Qianyi Xu, Zhongcong Yi, Xuanyu Yu, Zihao Zhang, Jianfeng Zheng, Zhuolin Zhu, Yifan Chen, Rui Guo, Hengkai Guo, Xiaoyang Han, Mingcong Han, Xu Li, Xiu Liang, Yixun Lou, Weiqiang Lu, Junzhe Luo, Guan Qin, Minghan Wang, Shuguang Wang, Yuang Graphics Computer Vision and Pattern Recognition Image and Video Processing We present Seed3D 2.0, an advanced 3D content generation system built on Seed3D 1.0, with substantial improvements across generation fidelity, simulation-ready capabilities, and application coverage. For geometry, a coarse-to-fine two-stage pipeline decouples global structure learning from high-frequency detail recovery, while a locality-aware VAE achieves higher spatial compression and more efficient decoding. For texture and material generation, we replace the cascaded pipeline of Seed3D 1.0 with a unified PBR model that directly generates multi-view albedo and metallic-roughness maps, enhanced by Mixture-of-Experts scaling and VLM-based semantic conditioning for improved material precision and visual fidelity. Beyond single-object generation, Seed3D 2.0 introduces a simulation-ready model suite comprising scene layout planning, part-aware decomposition, and training-free articulation generation, enabling coherent scene construction and part-level physical interaction across physics and graphics engines. A large-scale human preference study against five recent commercial models shows that Seed3D 2.0 achieves consistent win rates of 69.0% to 89.9% in textured 3D asset generation. Seed3D 2.0 is available on https://exp.volcengine.com/ark/vision?_vtm_=0.0.c70961.d701978.0&mode=vision&modelId=doubao-seed3d-2-0-260328&tab=Gen3D |
| title | Seed3D 2.0: Advancing High-Fidelity Simulation-Ready 3D Content Generation |
| topic | Graphics Computer Vision and Pattern Recognition Image and Video Processing |
| url | https://arxiv.org/abs/2605.13862 |