Image-Text-to-Video
Diffusers
Safetensors
English
Chinese
xgen-jing
world-model
egocentric
video-generation
audio-video-generation
Instructions to use XGENlabs/XGEN-JING with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use XGENlabs/XGEN-JING with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("XGENlabs/XGEN-JING", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
fix: remove unintended limitations and clean up prompt skills
Browse files
README.md
CHANGED
|
@@ -112,6 +112,12 @@ python3 demo_bidirection.py check_config=true
|
|
| 112 |
bash demo.sh cases=examples/bakery_greeting.json
|
| 113 |
```
|
| 114 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 115 |
Results are saved to `examples/outputs/`.
|
| 116 |
|
| 117 |
<details>
|
|
@@ -141,8 +147,9 @@ control entry per repeated slice:
|
|
| 141 |
}
|
| 142 |
```
|
| 143 |
|
| 144 |
-
`w/s/a/d` control forward/backward/left/right movement
|
| 145 |
-
|
|
|
|
| 146 |
With the default layout, `num_frames = 17 * sum(repeat) + 5`. Reference images
|
| 147 |
are ordered and addressed as `<Picture 1>`, `<Picture 2>`, and so on, up to five.
|
| 148 |
|
|
|
|
| 112 |
bash demo.sh cases=examples/bakery_greeting.json
|
| 113 |
```
|
| 114 |
|
| 115 |
+
Or:
|
| 116 |
+
|
| 117 |
+
```bash
|
| 118 |
+
bash demo.sh cases=examples/train_carriage_gaze.json
|
| 119 |
+
```
|
| 120 |
+
|
| 121 |
Results are saved to `examples/outputs/`.
|
| 122 |
|
| 123 |
<details>
|
|
|
|
| 147 |
}
|
| 148 |
```
|
| 149 |
|
| 150 |
+
`w/s/a/d` control forward/backward/left/right movement; `i/k` look up/down
|
| 151 |
+
and `j/l` turn left/right. `"w,a"` or `"w,j"` combines keys in one slice.
|
| 152 |
+
Opposite keys cancel; `""` applies no keys. Keep `control` the same length as `repeat`.
|
| 153 |
With the default layout, `num_frames = 17 * sum(repeat) + 5`. Reference images
|
| 154 |
are ordered and addressed as `<Picture 1>`, `<Picture 2>`, and so on, up to five.
|
| 155 |
|