Closed CiaraStrawberry closed 8 months ago
As you can see here: https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_video_diffusion/pipeline_stable_video_diffusion.py
during the forward pass, the init image shouldn't be scaled by the vae.config.scaling_factor, only the latents
Thank you very much for pointing out the mistake, I have already merged the change.
As you can see here: https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_video_diffusion/pipeline_stable_video_diffusion.py
during the forward pass, the init image shouldn't be scaled by the vae.config.scaling_factor, only the latents