Spaces:
Running
on
L40S
Running
on
L40S
Upload ./README_zh_cn.md with huggingface_hub
Browse files- README_zh_cn.md +38 -10
README_zh_cn.md
CHANGED
|
@@ -19,6 +19,9 @@
|
|
| 19 |
|
| 20 |
## 🔥🔥🔥 更新!!
|
| 21 |
|
|
|
|
|
|
|
|
|
|
| 22 |
* Nov 5, 2024: 💬 已经支持图生3D。请在[script](#using-gradio)体验。
|
| 23 |
* Nov 5, 2024: 💬 已经支持文生3D,请在[script](#using-gradio)体验。
|
| 24 |
|
|
@@ -27,9 +30,9 @@
|
|
| 27 |
|
| 28 |
- [x] Inference
|
| 29 |
- [x] Checkpoints
|
| 30 |
-
- [
|
| 31 |
-
- [ ] Training
|
| 32 |
- [ ] ComfyUI
|
|
|
|
| 33 |
- [ ] Distillation Version
|
| 34 |
- [ ] TensorRT Version
|
| 35 |
|
|
@@ -75,18 +78,16 @@ cd Hunyuan3D-1
|
|
| 75 |
env_install.sh 脚本提供了如何安装环境:
|
| 76 |
|
| 77 |
```
|
| 78 |
-
# 第一步:创建环境
|
| 79 |
conda create -n hunyuan3d-1 python=3.9 or 3.10 or 3.11 or 3.12
|
| 80 |
conda activate hunyuan3d-1
|
| 81 |
|
| 82 |
-
|
| 83 |
-
which pip # check pip corresponds to python
|
| 84 |
-
|
| 85 |
-
# modify the cuda version according to your machine (recommended)
|
| 86 |
-
pip install torch torchvision --index-url https://download.pytorch.org/whl/cu121
|
| 87 |
-
|
| 88 |
-
# 第三步:安装其他相关依赖包
|
| 89 |
bash env_install.sh
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 90 |
```
|
| 91 |
|
| 92 |
由于dust3r的许可证限制, 我们仅提供其安装途径:
|
|
@@ -204,6 +205,33 @@ bash scripts/image_to_3d_std_separately.sh ./demos/example_000.png ./outputs/tes
|
|
| 204 |
bash scripts/image_to_3d_lite_separately.sh ./demos/example_000.png ./outputs/test # >= 10G
|
| 205 |
```
|
| 206 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 207 |
#### Gradio界面部署
|
| 208 |
|
| 209 |
我们分别提供轻量版和标准版界面:
|
|
|
|
| 19 |
|
| 20 |
## 🔥🔥🔥 更新!!
|
| 21 |
|
| 22 |
+
* Nov 21, 2024: 💬 我们上传了新的纹理烘焙模块!
|
| 23 |
+
* Nov 20, 2024: 💬 我们添加了中文版的 README。
|
| 24 |
+
* Nov 18, 2024: 💬 感谢第三方开发者实现ComfyUI![[1]](https://github.com/jtydhr88/ComfyUI-Hunyuan3D-1-wrapper)[[2]](https://github.com/MrForExample/ComfyUI-3D-Pack)[[3]](https://github.com/TTPlanetPig/Comfyui_Hunyuan3D)
|
| 25 |
* Nov 5, 2024: 💬 已经支持图生3D。请在[script](#using-gradio)体验。
|
| 26 |
* Nov 5, 2024: 💬 已经支持文生3D,请在[script](#using-gradio)体验。
|
| 27 |
|
|
|
|
| 30 |
|
| 31 |
- [x] Inference
|
| 32 |
- [x] Checkpoints
|
| 33 |
+
- [x] Baking
|
|
|
|
| 34 |
- [ ] ComfyUI
|
| 35 |
+
- [ ] Training
|
| 36 |
- [ ] Distillation Version
|
| 37 |
- [ ] TensorRT Version
|
| 38 |
|
|
|
|
| 78 |
env_install.sh 脚本提供了如何安装环境:
|
| 79 |
|
| 80 |
```
|
|
|
|
| 81 |
conda create -n hunyuan3d-1 python=3.9 or 3.10 or 3.11 or 3.12
|
| 82 |
conda activate hunyuan3d-1
|
| 83 |
|
| 84 |
+
pip3 install torch torchvision --index-url https://download.pytorch.org/whl/cu121
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 85 |
bash env_install.sh
|
| 86 |
+
|
| 87 |
+
# or
|
| 88 |
+
pip3 install -r requirements.txt --index-url https://download.pytorch.org/whl/cu121
|
| 89 |
+
pip3 install git+https://github.com/facebookresearch/pytorch3d@stable
|
| 90 |
+
pip3 install git+https://github.com/NVlabs/nvdiffrast
|
| 91 |
```
|
| 92 |
|
| 93 |
由于dust3r的许可证限制, 我们仅提供其安装途径:
|
|
|
|
| 205 |
bash scripts/image_to_3d_lite_separately.sh ./demos/example_000.png ./outputs/test # >= 10G
|
| 206 |
```
|
| 207 |
|
| 208 |
+
#### 纹理烘焙
|
| 209 |
+
|
| 210 |
+
我们提供了纹理烘焙模块。对齐和变形过程是使用Dust3R完成的,遵守CC BY-NC-SA 4.0许可。请注意,这是一个非商业许可证,因此该模块不能用于商业目的。
|
| 211 |
+
|
| 212 |
+
```bash
|
| 213 |
+
mkdir -p ./third_party/weights/DUSt3R_ViTLarge_BaseDecoder_512_dpt
|
| 214 |
+
huggingface-cli download naver/DUSt3R_ViTLarge_BaseDecoder_512_dpt \
|
| 215 |
+
--local-dir ./third_party/weights/DUSt3R_ViTLarge_BaseDecoder_512_dpt
|
| 216 |
+
|
| 217 |
+
cd ./third_party
|
| 218 |
+
git clone --recursive https://github.com/naver/dust3r.git
|
| 219 |
+
|
| 220 |
+
cd ..
|
| 221 |
+
```
|
| 222 |
+
如果您使用相关代码和权重,我们也列出一些烘焙相关参数:
|
| 223 |
+
|
| 224 |
+
| Argument | Default | Description |
|
| 225 |
+
|:------------------:|:---------:|:---------------------------------------------------:|
|
| 226 |
+
|`--do_bake` | False | baking multi-view images onto mesh |
|
| 227 |
+
|`--bake_align_times` | 3 | alignment number of image and mesh |
|
| 228 |
+
|
| 229 |
+
|
| 230 |
+
注意:如果需要烘焙,请确保`--do_bake`设置为`True`并且`--do_texture_mapping`也设置为`True`。
|
| 231 |
+
|
| 232 |
+
```bash
|
| 233 |
+
python main.py ... --do_texture_mapping --do_bake (--do_render)
|
| 234 |
+
|
| 235 |
#### Gradio界面部署
|
| 236 |
|
| 237 |
我们分别提供轻量版和标准版界面:
|