Spaces:

multimodalart

/

stable-video-diffusion

like 1.15k

Running on A10G

not working indeed

Hi it is working but the queue can get full. You can keep trying to get in

It's not about the queue here. This is simply an "error" message.

Please fix! I have tried with many different images and parameters, always gives error :(
Also could not get this to work in my own space, always got out memory error with any node type.

Hello!

We've made a couple of changes to reduce memory use, so it should be more stable now. We've also improved error reporting so in most cases you should see a description when an error occurs. Please, do let us know if you experience issues again.

Do keep in mind that this is a very compute-intensive space, so:

• The queue sometimes is full and doesn't accept more requests. We recommend you wait a few minutes before retrying.
• Inference time is long, it takes more than a minute to generate a video.
• Even if you duplicate the Space, it won't work on A10 GPU out of the box as it doesn't have enough GPU memory. This demo requires an A100 for now. We are investigating how to remove this limitation.

Thanks all for your input!

works well!