Update README.md
Browse files
README.md
CHANGED
|
@@ -1,103 +1,104 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
-
|
| 12 |
-
-
|
| 13 |
-
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
- [ ]
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
<td>
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
<td><img src="
|
| 29 |
-
|
| 30 |
-
</
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
<td>
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
<td><img src="
|
| 40 |
-
|
| 41 |
-
</
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
<td>
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
<td><img src="
|
| 51 |
-
|
| 52 |
-
</
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
<td>
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
<td><img src="
|
| 62 |
-
|
| 63 |
-
</
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
<td>
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
<td><img src="
|
| 73 |
-
|
| 74 |
-
</
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
mkdir -p models/
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
|
| 102 |
-
|
|
|
|
| 103 |
Then run the file `examples/z_image_fun/predict_t2i_control.py`.
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: videox_fun
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
# Z-Image-Turbo-Fun-Controlnet-Union
|
| 7 |
+
|
| 8 |
+
[](https://github.com/aigc-apps/VideoX-Fun)
|
| 9 |
+
|
| 10 |
+
## Model Features
|
| 11 |
+
- This ControlNet is added on 6 blocks.
|
| 12 |
+
- The model was trained from scratch for 10,000 steps on a dataset of 1 million high-quality images covering both general and human-centric content. Training was performed at 1328 resolution using BFloat16 precision, with a batch size of 64, a learning rate of 2e-5, and a text dropout ratio of 0.10.
|
| 13 |
+
- It supports multiple control conditions—including Canny, HED, Depth, Pose and MLSD can be used like a standard ControlNet.
|
| 14 |
+
- You can adjust control_context_scale for stronger control and better detail preservation. For better stability, we highly recommend using a detailed prompt. The optimal range for control_context_scale is from 0.65 to 0.80.
|
| 15 |
+
|
| 16 |
+
## TODO
|
| 17 |
+
- [ ] Train on more data and for more steps.
|
| 18 |
+
- [ ] Support inpaint mode.
|
| 19 |
+
|
| 20 |
+
## Results
|
| 21 |
+
|
| 22 |
+
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
| 23 |
+
<tr>
|
| 24 |
+
<td>Pose</td>
|
| 25 |
+
<td>Output</td>
|
| 26 |
+
</tr>
|
| 27 |
+
<tr>
|
| 28 |
+
<td><img src="asset/pose2.jpg" width="100%" /></td>
|
| 29 |
+
<td><img src="results/pose2.png" width="100%" /></td>
|
| 30 |
+
</tr>
|
| 31 |
+
</table>
|
| 32 |
+
|
| 33 |
+
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
| 34 |
+
<tr>
|
| 35 |
+
<td>Pose</td>
|
| 36 |
+
<td>Output</td>
|
| 37 |
+
</tr>
|
| 38 |
+
<tr>
|
| 39 |
+
<td><img src="asset/pose.jpg" width="100%" /></td>
|
| 40 |
+
<td><img src="results/pose.png" width="100%" /></td>
|
| 41 |
+
</tr>
|
| 42 |
+
</table>
|
| 43 |
+
|
| 44 |
+
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
| 45 |
+
<tr>
|
| 46 |
+
<td>Canny</td>
|
| 47 |
+
<td>Output</td>
|
| 48 |
+
</tr>
|
| 49 |
+
<tr>
|
| 50 |
+
<td><img src="asset/canny.jpg" width="100%" /></td>
|
| 51 |
+
<td><img src="results/canny.png" width="100%" /></td>
|
| 52 |
+
</tr>
|
| 53 |
+
</table>
|
| 54 |
+
|
| 55 |
+
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
| 56 |
+
<tr>
|
| 57 |
+
<td>HED</td>
|
| 58 |
+
<td>Output</td>
|
| 59 |
+
</tr>
|
| 60 |
+
<tr>
|
| 61 |
+
<td><img src="asset/hed.jpg" width="100%" /></td>
|
| 62 |
+
<td><img src="results/hed.png" width="100%" /></td>
|
| 63 |
+
</tr>
|
| 64 |
+
</table>
|
| 65 |
+
|
| 66 |
+
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
| 67 |
+
<tr>
|
| 68 |
+
<td>Depth</td>
|
| 69 |
+
<td>Output</td>
|
| 70 |
+
</tr>
|
| 71 |
+
<tr>
|
| 72 |
+
<td><img src="asset/depth.jpg" width="100%" /></td>
|
| 73 |
+
<td><img src="results/depth.png" width="100%" /></td>
|
| 74 |
+
</tr>
|
| 75 |
+
</table>
|
| 76 |
+
|
| 77 |
+
## Inference
|
| 78 |
+
Go to the VideoX-Fun repository for more details.
|
| 79 |
+
|
| 80 |
+
Please clone the VideoX-Fun repository and create the required directories:
|
| 81 |
+
|
| 82 |
+
```sh
|
| 83 |
+
# Clone the code
|
| 84 |
+
git clone https://github.com/aigc-apps/VideoX-Fun.git
|
| 85 |
+
|
| 86 |
+
# Enter VideoX-Fun's directory
|
| 87 |
+
cd VideoX-Fun
|
| 88 |
+
|
| 89 |
+
# Create model directories
|
| 90 |
+
mkdir -p models/Diffusion_Transformer
|
| 91 |
+
mkdir -p models/Personalized_Model
|
| 92 |
+
```
|
| 93 |
+
|
| 94 |
+
Then download the weights into models/Diffusion_Transformer and models/Personalized_Model.
|
| 95 |
+
|
| 96 |
+
```
|
| 97 |
+
📦 models/
|
| 98 |
+
├── 📂 Diffusion_Transformer/
|
| 99 |
+
│ └── 📂 Z-Image-Turbo/
|
| 100 |
+
├── 📂 Personalized_Model/
|
| 101 |
+
│ └── 📦 Z-Image-Turbo-Fun-Controlnet-Union.safetensors
|
| 102 |
+
```
|
| 103 |
+
|
| 104 |
Then run the file `examples/z_image_fun/predict_t2i_control.py`.
|