Image-to-3D
Yang2001 nielsr HF Staff commited on
Commit
7bde14f
·
1 Parent(s): 467a6a2

Add image-to-3d pipeline tag and improve model card (#1)

Browse files

- Add image-to-3d pipeline tag and improve model card (45fafc48f156a48954cb4d3a2805e9df39c564a1)


Co-authored-by: Niels Rogge <nielsr@users.noreply.huggingface.co>

Files changed (1) hide show
  1. README.md +108 -117
README.md CHANGED
@@ -1,117 +1,108 @@
1
- ---
2
- license: other
3
- license_name: pixal3d-license
4
- license_link: LICENSE
5
- extra_gated_eu_disallowed: true
6
- ---
7
-
8
- <div align="center">
9
-
10
- # Pixal3D: Pixel-Aligned 3D Generation from Images
11
-
12
- <h3>SIGGRAPH 2026</h3>
13
-
14
- [Dong-Yang Li](https://ldyang694.github.io/)¹ · [Wang Zhao](https://thuzhaowang.github.io/)²* · [Yuxin Chen](https://orcid.org/0000-0002-7854-1072)² · [Wenbo Hu](https://wbhu.github.io/)² · [Meng-Hao Guo](https://menghaoguo.github.io/)¹ · [Fang-Lue Zhang](https://fanglue.github.io/)³ · [Ying Shan](https://www.linkedin.com/in/YingShanProfile)² · [Shi-Min Hu](https://cg.cs.tsinghua.edu.cn/shimin.htm)¹✉
15
-
16
- ¹Tsinghua University (BNRist) &nbsp;&nbsp; ²Tencent ARC Lab &nbsp;&nbsp; ³Victoria University of Wellington
17
-
18
- *Project lead &nbsp;&nbsp; ✉Corresponding author
19
-
20
- </div>
21
-
22
- <div align="center">
23
- <a href="https://ldyang694.github.io/projects/pixal3d/"><img src=https://img.shields.io/badge/Project%20Page-333399.svg?logo=googlehome height=22px></a>
24
- <a href="https://huggingface.co/spaces/TencentARC/Pixal3D"><img src=https://img.shields.io/badge/%F0%9F%A4%97%20Demo-276cb4.svg height=22px></a>
25
- <a href="https://huggingface.co/TencentARC/Pixal3D"><img src=https://img.shields.io/badge/%F0%9F%A4%97%20Models-d96902.svg height=22px></a>
26
- <a href="https://arxiv.org/abs/2605.10922"><img src=https://img.shields.io/badge/Arxiv-b5212f.svg?logo=arxiv height=22px></a>
27
- </div>
28
-
29
- **Pixal3D** generates high-fidelity 3D assets from a single image. Unlike previous methods that loosely inject image features via attention, Pixal3D explicitly lifts pixel features into 3D through back-projection, establishing direct pixel-to-3D correspondences. This enables near-reconstruction-level fidelity with detailed geometry and PBR textures.
30
-
31
- ---
32
-
33
- ## ✨ News
34
-
35
- - **May 2026**: Release the improved version based on [Trellis.2](https://github.com/microsoft/TRELLIS.2) backbone. 💪
36
- - **May 2026**: Release inference code and online demo. 🤗
37
- - **Apr 2026**: Our paper is accepted to SIGGRAPH 2026! 🎉
38
-
39
- ## 📌 Branches
40
-
41
- | Branch | Description |
42
- |--------|-------------|
43
- | `main` | **Latest version** — improved implementation based on [Trellis.2](https://github.com/microsoft/TRELLIS.2) backbone with better performance. |
44
- | `paper` | **Paper version** — original implementation based on [Direct3D-S2](https://github.com/DreamTechAI/Direct3D-S2), corresponding to results reported in our SIGGRAPH 2026 paper. |
45
-
46
- > If you want to reproduce the results in our paper, please switch to the `paper` branch.
47
-
48
- ## 🎮 Try It Online
49
-
50
- You can try Pixal3D directly in your browser without any installation via our Hugging Face Gradio demo:
51
-
52
- 👉 [**Launch Demo**](https://huggingface.co/spaces/TencentARC/Pixal3D)
53
-
54
- ## 🚀 Getting Started
55
-
56
- ### Installation
57
-
58
- #### Step 1: Follow TRELLIS.2 Installation
59
-
60
- Please first follow the installation guide of [TRELLIS.2](https://github.com/microsoft/TRELLIS.2) to set up the base environment.
61
-
62
- #### Step 2: Install Additional Dependencies
63
-
64
- ```bash
65
- pip install -r requirements.txt
66
- ```
67
-
68
- #### Step 3: Install utils3d
69
-
70
- ```bash
71
- pip install https://github.com/LDYang694/Storages/releases/download/20260430/utils3d-0.0.2-py3-none-any.whl
72
- ```
73
-
74
- > **Note**: `requirements-hfdemo.txt` is for the Hugging Face Spaces demo (H-series GPU architecture) and may not be compatible with other architectures.
75
-
76
- ### Usage
77
-
78
- #### Inference
79
-
80
- Generate a GLB mesh from a single image:
81
-
82
- ```bash
83
- python inference.py --image assets/test_image/0.png --output ./output.glb
84
- ```
85
-
86
- ### Web Demo
87
-
88
- We provide a Gradio web demo for Pixal3D, which allows you to generate 3D meshes from images interactively.
89
-
90
- ```bash
91
- python app.py
92
- ```
93
-
94
- ## 🤗 Acknowledgements
95
-
96
- This project is heavily built upon [Trellis.2](https://github.com/microsoft/TRELLIS.2) and [Direct3D-S2](https://github.com/DreamTechAI/Direct3D-S2). We sincerely thank the authors for their outstanding work on scalable 3D generation , which serves as the foundation of our codebase and model architecture.
97
-
98
- We also thank the following repos for their great contributions:
99
-
100
- - [Direct3D-S2](https://github.com/DreamTechAI/Direct3D-S2)
101
- - [Trellis](https://github.com/microsoft/TRELLIS)
102
- - [Trellis.2](https://github.com/microsoft/TRELLIS.2)
103
-
104
- ## 📄 Citation
105
-
106
- If you find this work useful, please consider citing:
107
-
108
- ```bibtex
109
- @article{li2026pixal3d,
110
- title = {Pixal3D: Pixel-Aligned 3D Generation from Images},
111
- author = {Li, Dong-Yang and Zhao, Wang and Chen, Yuxin and Hu, Wenbo and Guo, Meng-Hao and Zhang, Fang-Lue and Shan, Ying and Hu, Shi-Min},
112
- journal = {arXiv preprint arXiv:2605.10922},
113
- year = {2026}
114
- }
115
- ```
116
-
117
-
 
1
+ ---
2
+ license: other
3
+ license_name: pixal3d-license
4
+ license_link: LICENSE
5
+ extra_gated_eu_disallowed: true
6
+ pipeline_tag: image-to-3d
7
+ ---
8
+
9
+ <div align="center">
10
+
11
+ # Pixal3D: Pixel-Aligned 3D Generation from Images
12
+
13
+ <h3>SIGGRAPH 2026</h3>
14
+
15
+ [Dong-Yang Li](https://ldyang694.github.io/)¹ · [Wang Zhao](https://thuzhaowang.github.io/)²* · [Yuxin Chen](https://orcid.org/0000-0002-7854-1072)² · [Wenbo Hu](https://wbhu.github.io/)² · [Meng-Hao Guo](https://menghaoguo.github.io/)¹ · [Fang-Lue Zhang](https://fanglue.github.io/)³ · [Ying Shan](https://www.linkedin.com/in/YingShanProfile)² · [Shi-Min Hu](https://cg.cs.tsinghua.edu.cn/shimin.htm)¹✉
16
+
17
+ ¹Tsinghua University (BNRist) &nbsp;&nbsp; ²Tencent ARC Lab &nbsp;&nbsp; ³Victoria University of Wellington
18
+
19
+ *Project lead &nbsp;&nbsp; ✉Corresponding author
20
+
21
+ </div>
22
+
23
+ <div align="center">
24
+ <a href="https://ldyang694.github.io/projects/pixal3d/"><img src=https://img.shields.io/badge/Project%20Page-333399.svg?logo=googlehome height=22px></a>
25
+ <a href="https://huggingface.co/spaces/TencentARC/Pixal3D"><img src=https://img.shields.io/badge/%F0%9F%A4%97%20Demo-276cb4.svg height=22px></a>
26
+ <a href="https://github.com/TencentARC/Pixal3D"><img src=https://img.shields.io/badge/Code-Github-black.svg?logo=github height=22px></a>
27
+ <a href="https://arxiv.org/abs/2605.10922"><img src=https://img.shields.io/badge/Arxiv-b5212f.svg?logo=arxiv height=22px></a>
28
+ </div>
29
+
30
+ **Pixal3D** generates high-fidelity 3D assets from a single image. Unlike previous methods that loosely inject image features via attention, Pixal3D explicitly lifts pixel features into 3D through back-projection, establishing direct pixel-to-3D correspondences. This enables near-reconstruction-level fidelity with detailed geometry and PBR textures.
31
+
32
+ ---
33
+
34
+ ## ✨ News
35
+
36
+ - **May 2026**: Release the improved version based on [Trellis.2](https://github.com/microsoft/TRELLIS.2) backbone. 💪
37
+ - **May 2026**: Release inference code and online demo. 🤗
38
+ - **Apr 2026**: Our paper is accepted to SIGGRAPH 2026! 🎉
39
+
40
+ ## 📌 Branches
41
+
42
+ | Branch | Description |
43
+ |--------|-------------|
44
+ | `main` | **Latest version** — improved implementation based on [Trellis.2](https://github.com/microsoft/TRELLIS.2) backbone with better performance. |
45
+ | `paper` | **Paper version** — original implementation based on [Direct3D-S2](https://github.com/DreamTechAI/Direct3D-S2), corresponding to results reported in our SIGGRAPH 2026 paper. |
46
+
47
+ > If you want to reproduce the results in our paper, please switch to the `paper` branch.
48
+
49
+ ## 🎮 Try It Online
50
+
51
+ You can try Pixal3D directly in your browser without any installation via our Hugging Face Gradio demo:
52
+
53
+ 👉 [**Launch Demo**](https://huggingface.co/spaces/TencentARC/Pixal3D)
54
+
55
+ ## 🚀 Getting Started
56
+
57
+ ### Installation
58
+
59
+ #### Step 1: Follow TRELLIS.2 Installation
60
+
61
+ Please first follow the installation guide of [TRELLIS.2](https://github.com/microsoft/TRELLIS.2) to set up the base environment.
62
+
63
+ #### Step 2: Install Additional Dependencies
64
+
65
+ ```bash
66
+ pip install -r requirements.txt
67
+ ```
68
+
69
+ #### Step 3: Install utils3d
70
+
71
+ ```bash
72
+ pip install https://github.com/LDYang694/Storages/releases/download/20260430/utils3d-0.0.2-py3-none-any.whl
73
+ ```
74
+
75
+ ### Usage
76
+
77
+ #### Inference
78
+
79
+ Generate a GLB mesh from a single image:
80
+
81
+ ```bash
82
+ python inference.py --image assets/test_image/0.png --output ./output.glb
83
+ ```
84
+
85
+ ### Web Demo
86
+
87
+ We provide a Gradio web demo for Pixal3D, which allows you to generate 3D meshes from images interactively.
88
+
89
+ ```bash
90
+ python app.py
91
+ ```
92
+
93
+ ## 🤗 Acknowledgements
94
+
95
+ This project is heavily built upon [Trellis.2](https://github.com/microsoft/TRELLIS.2) and [Direct3D-S2](https://github.com/DreamTechAI/Direct3D-S2). We also thank the following repos for their great contributions: [Trellis](https://github.com/microsoft/TRELLIS).
96
+
97
+ ## 📄 Citation
98
+
99
+ If you find this work useful, please consider citing:
100
+
101
+ ```bibtex
102
+ @article{li2026pixal3d,
103
+ title = {Pixal3D: Pixel-Aligned 3D Generation from Images},
104
+ author = {Li, Dong-Yang and Zhao, Wang and Chen, Yuxin and Hu, Wenbo and Guo, Meng-Hao and Zhang, Fang-Lue and Shan, Ying and Hu, Shi-Min},
105
+ journal = {arXiv preprint arXiv:2605.10922},
106
+ year = {2026}
107
+ }
108
+ ```