Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

About creating motion Lora #8

Open
AA-Developer opened this issue Aug 22, 2024 · 1 comment
Open

About creating motion Lora #8

AA-Developer opened this issue Aug 22, 2024 · 1 comment

Comments

@AA-Developer
Copy link

Hi
When will the code for creating a motion Lora that works with your system be published?
I think it is possible to create motion Lora using the old Animatediff But I don't think the result will be not good because.
Also is your model compatible with finetune ?
Also Are you planning to release an improved version of the large model (SD XL) ?

@MaAo
Copy link
Collaborator

MaAo commented Aug 23, 2024

Hi When will the code for creating a motion Lora that works with your system be published? I think it is possible to create motion Lora using the old Animatediff But I don't think the result will be not good because. Also is your model compatible with finetune ? Also Are you planning to release an improved version of the large model (SD XL) ?

Thank you for your interest in our work. As outlined in our paper, we did not use motion Lora for motion control. Instead, we computed the motion score from each training video with RAFT and injected this score into our UNet for adjustment. Thus, during inference, users can input the motion score to control the amplitude of the generated video's motion.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants