volotat / SD-CN-Animation

This script allows to automate video stylization task using StableDiffusion and ControlNet.
MIT License
807 stars 62 forks source link

Elaborate on the meaning of input settings #113

Open aiXander opened 1 year ago

aiXander commented 1 year ago

Absolutely loving this!! Could you please add some information about what the various input settings do to the README?

volotat commented 1 year ago

There is a discussion here https://github.com/volotat/SD-CN-Animation/discussions/43 about these parameters. I think I could add links to this discussion in the readme. But I really don't want to overload readme with too much information and present only the most important stuff.

lauretta91 commented 1 year ago

Hi volotat, I see the Processing strength and Fix frame strength explanation, I dont see the explanation for the Occlusion mask params. Could you please link us to where we can find it pleeeease? Thanks :)

aulerius commented 1 year ago

I'd also be very interested in explanations for vid2vid specific advanced settings, regarding consistency, occlusion, etc. I get the jist of it and the principles behind it, but can't quite confidently imagine the exact affects of each parameter isolated.

Enferlain commented 1 year ago

I'm interested in the occlusion settings as well

lauretta91 commented 1 year ago

@volotat I really like this extension, and thank you!

These advanced settings are linked to the RAFT method, I suppose, but it's very difficult to understand the impact, given the generation process is very slow. Volotat, is the speed something you are looking into?