Image-to-3D
nielsr HF Staff commited on
Commit
45fafc4
·
verified ·
1 Parent(s): 467a6a2

Add image-to-3d pipeline tag and improve model card

Browse files

Hi, I'm Niels from the Hugging Face community science team.

I've opened this PR to add the `pipeline_tag: image-to-3d` to your model card. This metadata is essential for ensuring your model is correctly indexed and discoverable by users interested in 3D generation on the Hugging Face Hub.

I have also updated the model card to include the "Getting Started" and "Usage" sections from your GitHub README, making it easier for users to understand how to run the model.

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
+ ```