File size: 2,206 Bytes
4711b5a 7fd7827 4711b5a 7fd7827 4711b5a b9dbf9f 4711b5a 367dd3c 4711b5a 493a3d9 4711b5a 493a3d9 4711b5a 493a3d9 4711b5a 493a3d9 b9dbf9f 493a3d9 4711b5a 97cb3c0 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 |
---
tags:
- Text-to-Video
license: cc-by-nc-4.0
---
# zeroscope_v2 30x448x256
A watermark-free Modelscope-based video model optimized for producing high-quality 16:9 compositions and a smooth video output. This model was trained from the [original weights](https://huggingface.co/damo-vilab/modelscope-damo-text-to-video-synthesis) using 9,923 clips and 29,769 tagged frames at 30 frames, 448x256 resolution.<br />
zeroscope_v2 30x448x256 is specifically designed for upscaling with [Potat1](https://huggingface.co/camenduru/potat1) using vid2vid in the [1111 text2video](https://github.com/kabachuha/sd-webui-text2video) extension by [kabachuha](https://github.com/kabachuha). Leveraging this model as a preliminary step allows for superior overall compositions at higher resolutions in Potat1, permitting faster exploration in 448x256 before transitioning to a high-resolution render. See an [example output](https://i.imgur.com/lj90FYP.mp4) that has been upscaled to 1152 x 640 using Potat1.<br />
### Using it with the 1111 text2video extension
1. Rename the file 'zeroscope_v2_30x448x256.pth' to 'text2video_pytorch_model.pth'.
2. Rename the file 'zeroscope_v2_30x448x256_text.bin' to 'open_clip_pytorch_model.bin'.
3. Replace the respective files in the 'stable-diffusion-webui\models\ModelScope\t2v' directory.
### Upscaling recommendations
For upscaling, it's recommended to use Potat1 via vid2vid in the 1111 extension. Aim for a resolution of 1152x640 and a denoise strength between 0.66 and 0.85. Remember to use the same prompt and settings that were used to generate the original clip.
### Known issues
Lower resolutions or fewer frames could lead to suboptimal output. <br />
Certain clips might appear with cuts. This will be fixed in the upcoming 2.1 version, which will incorporate a cleaner dataset.
Some clips may playback too slowly, requiring prompt engineering for an increased pace.
Thanks to [camenduru](https://github.com/camenduru), [kabachuha](https://github.com/kabachuha), [ExponentialML](https://github.com/ExponentialML), [polyware](https://twitter.com/polyware_ai), [tin2tin](https://github.com/tin2tin)<br /> |