Upload folder using huggingface_hub
Browse files- .gitattributes +1 -0
- README.md +7 -7
- dreamx-creator_teaser.png +3 -0
.gitattributes
CHANGED
|
@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
wan2.2_ti2v_5b/google/umt5-xxl/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
wan2.2_ti2v_5b/google/umt5-xxl/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
dreamx-creator_teaser.png filter=lfs diff=lfs merge=lfs -text
|
README.md
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
<div align="center">
|
| 2 |
-
<img src=".
|
| 3 |
|
| 4 |
<h1>DreamX-Creator 1.0: Model Weights</h1>
|
| 5 |
|
|
@@ -28,7 +28,7 @@ following the layout below.
|
|
| 28 |
|
| 29 |
<!-- TODO: fill in the HuggingFace / ModelScope links above -->
|
| 30 |
|
| 31 |
-
##
|
| 32 |
|
| 33 |
```
|
| 34 |
checkpoints/
|
|
@@ -52,7 +52,7 @@ The `wan2.2_ti2v_5b/` directory can also be downloaded directly from
|
|
| 52 |
[Wan-AI/Wan2.2-TI2V-5B](https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B); only the
|
| 53 |
three entries above are needed. It is shared by both pipelines.
|
| 54 |
|
| 55 |
-
##
|
| 56 |
|
| 57 |
- **Joint audio-video generation** uses `creator/`, `audio_vae/`, and
|
| 58 |
`wan2.2_ti2v_5b/`. See the
|
|
@@ -62,7 +62,7 @@ three entries above are needed. It is shared by both pipelines.
|
|
| 62 |
[video_refiner README](../video_refiner/README.md) for setup and the full
|
| 63 |
list of inference knobs.
|
| 64 |
|
| 65 |
-
##
|
| 66 |
|
| 67 |
Once the weights above are in place, from the repository root:
|
| 68 |
|
|
@@ -84,7 +84,7 @@ pip install -r requirements.txt
|
|
| 84 |
INPUT=/path/to/video.mp4 bash run_inference.sh
|
| 85 |
```
|
| 86 |
|
| 87 |
-
##
|
| 88 |
|
| 89 |
If you find DreamX-Creator useful in your research, please consider citing our technical report:
|
| 90 |
|
|
@@ -100,10 +100,10 @@ If you find DreamX-Creator useful in your research, please consider citing our t
|
|
| 100 |
}
|
| 101 |
```
|
| 102 |
|
| 103 |
-
##
|
| 104 |
|
| 105 |
This project is licensed under the Apache License 2.0. See [LICENSE](../LICENSE) for details.
|
| 106 |
|
| 107 |
-
##
|
| 108 |
|
| 109 |
We would like to thank the [Wan Team](https://github.com/Wan-Video/Wan2.2), the [OpenMOSS Team](https://github.com/OpenMOSS/MOVA), and the [VideoX-Fun Team](https://github.com/aigc-apps/VideoX-Fun) for their outstanding open-source work on Wan, MOVA, and VideoX-Fun, respectively.
|
|
|
|
| 1 |
<div align="center">
|
| 2 |
+
<img src="./dreamx-creator_teaser.png" alt="DreamX-Creator teaser">
|
| 3 |
|
| 4 |
<h1>DreamX-Creator 1.0: Model Weights</h1>
|
| 5 |
|
|
|
|
| 28 |
|
| 29 |
<!-- TODO: fill in the HuggingFace / ModelScope links above -->
|
| 30 |
|
| 31 |
+
## Expected Layout
|
| 32 |
|
| 33 |
```
|
| 34 |
checkpoints/
|
|
|
|
| 52 |
[Wan-AI/Wan2.2-TI2V-5B](https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B); only the
|
| 53 |
three entries above are needed. It is shared by both pipelines.
|
| 54 |
|
| 55 |
+
## Which Weights Are Used Where
|
| 56 |
|
| 57 |
- **Joint audio-video generation** uses `creator/`, `audio_vae/`, and
|
| 58 |
`wan2.2_ti2v_5b/`. See the
|
|
|
|
| 62 |
[video_refiner README](../video_refiner/README.md) for setup and the full
|
| 63 |
list of inference knobs.
|
| 64 |
|
| 65 |
+
## Quickstart
|
| 66 |
|
| 67 |
Once the weights above are in place, from the repository root:
|
| 68 |
|
|
|
|
| 84 |
INPUT=/path/to/video.mp4 bash run_inference.sh
|
| 85 |
```
|
| 86 |
|
| 87 |
+
## Citation
|
| 88 |
|
| 89 |
If you find DreamX-Creator useful in your research, please consider citing our technical report:
|
| 90 |
|
|
|
|
| 100 |
}
|
| 101 |
```
|
| 102 |
|
| 103 |
+
## License
|
| 104 |
|
| 105 |
This project is licensed under the Apache License 2.0. See [LICENSE](../LICENSE) for details.
|
| 106 |
|
| 107 |
+
## Acknowledgement
|
| 108 |
|
| 109 |
We would like to thank the [Wan Team](https://github.com/Wan-Video/Wan2.2), the [OpenMOSS Team](https://github.com/OpenMOSS/MOVA), and the [VideoX-Fun Team](https://github.com/aigc-apps/VideoX-Fun) for their outstanding open-source work on Wan, MOVA, and VideoX-Fun, respectively.
|
dreamx-creator_teaser.png
ADDED
|
Git LFS Details
|