stable-diffusion-webui
stable-diffusion-webui copied to clipboard
AUTOMATIC1111
[Feature Request]: Implement Stable Video Diffusion model (SVD)
Is there an existing issue for this?
- [X] I have searched the existing issues and checked the recent builds/commits
What would your feature do?
Stability AI just dropped their new model Stable Video Diffusion Weights are already available for download. Here is the paper
It would be awesome to have it in the most popular SD UI
Proposed workflow
- Go to text-2-video tab
- Enter the prompt
- Click Generate
- Enjoy the show
Additional information
No response
Some useful tips and tricks on what's the optimal parameters from one of the developers: https://x.com/timudk/status/1727064128223855087?s=20
https://github.com/continue-revolution/sd-webui-animatediff/issues/339#issuecomment-1822207493
should rename to have SVD in title https://x.com/fofrAI/status/1727620324563251494?s=20
found this https://github.com/kijai/ComfyUI-SVD
https://github.com/replicate/cog-svd
should rename to have SVD in title https://x.com/fofrAI/status/1727620324563251494?s=20
found this https://github.com/kijai/ComfyUI-SVD
https://github.com/replicate/cog-svd
why do they only have 14 or 25 frames? i thought you can generate as much as you want? And im waiting on automatic1111 to have this implemented, i dont want to move around my checkpoints
continue-revolution have said that he won't be able to support this in sd-webui-animatediff until next week Wednesday at the earliest, no guarantees on the exact time as he's currently busy with all the matters see post https://github.com/continue-revolution/sd-webui-animatediff/issues/351
please stop the +1 spam
Boom 💥 https://github.com/xx025/stable-video-diffusion-webui
is this only img to video? can it run txt 2 vid?
continue-revolution have said that he won't be able to support this in sd-webui-animatediff until next week Wednesday at the earliest, no guarantees on the exact time as he's currently busy with all the matters see post continue-revolution/sd-webui-animatediff#351
please stop the
+1spam
We need official native support in a txt2video tab. Not crappy extensions.
@enriquemorenotent i'm interested in the feature, i want updates about it...not that someone wants it? +1 spam is annoying
@taybart So, if you are interested in the feature... why would you want the thread locked?
You want the feature, but the discussion of the feature is (understandably) annoying you. It is kind of a paradoxical situation. You might be better off bookmarking it, and unsubscribing.
@taybart So, if you are interested in the feature... why would you want the thread locked?
You want the feature, but the discussion of the feature is (understandably) annoying you. It is kind of a paradoxical situation. You might be better off bookmarking it, and unsubscribing.
People do it because developers ignore us. So it is better to bump it
+1 +1 +1
People do it because developers ignore us. So it is better to bump it
Opensource doesn't work this way. If you want this feature - you implement it and create PR. Developers don't owe anything to anybody because they are doing this for free. So, if you're not going to work on a feature, stop the flood in the issue please. There are more that 1700 other issues they need to address.
Be patient everybody pretty please.
People do it because developers ignore us. So it is better to bump it
People do it, because they want to show the interest in the feature. Even though arguably a :+1: in the opening post should be just as efficient to do that
People do it because developers ignore us. So it is better to bump it
People do it, because they want to show the interest in the feature. Even though arguably a 👍 in the opening post should be just as efficient to do that
Yes this is one of the most requested features now and it has been one month, and I see no progress.
People do it because developers ignore us. So it is better to bump it
People do it, because they want to show the interest in the feature. Even though arguably a 👍 in the opening post should be just as efficient to do that
Yes this is one of the most requested features now and it has been one month, and I see no progress.
We have the right to show interest, but not the right to demand the developers to do it.
Like others said, if you want it so badly... do it yourself :)
For all these new folks here. "+1" on GitHub is considered a very bad taste. As well as arguing ^_^ Let's keep discussions productive please.
sd-webui-forge fork have implementation of SVD. Maybe somebody will adapt it for original webui? https://github.com/lllyasviel/stable-diffusion-webui-forge/blob/main/extensions-builtin/sd_forge_svd/scripts/forge_svd.py
