-
Notifications
You must be signed in to change notification settings - Fork 189
Open
Description
In the standard WanFirstLastFrameToVideo node, the starting image input and and width/depth inputs are separate. So I can create a 720p video with a 1440p starting image, which gives a very good quality and crisp starting video.
In this SVI setup (which is awesome by the way), it uses the image's resolution as the sampling resolution, so I have to of course downscale 1440p to 720p to fit on my 5090's 32GB of VRAM. This makes the first video not very high quality anymore, as compared to the standard Wan node.
Is there any way possible with this SVI setup to have the starting image resolution, and the sampling resolution, separate from each other? There doesn't seem to be a way..
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels