How do you use custom models for videos?

[TeKett]

Videos use 2 models, a high noise and a low noise, or basically a base and refiner. But even for images i dont see how to select a refiner model, only another sampler. In A1111 you can select another base model to do the latter part of the inference. Videos do the same thing. But i dont see anywhere to select a second base model for the refiner.

I already have the base wan 2.2 model downloaded (from the comfy repacks), so i dont want to redownload it. Second, i want to use models from civit and loras.

[vladmandic]

you can choose if you run 1st stage only, 2nd stage only or both: settings -> model options -> wanai.

for video models, in general you must download them via sdnext - format is different than comfy repacks. it could be made to work, but it hasn't been highest priority.

loras you can use freely as usual.
for wan video we currently do not support 3rd party model finetunes, its just been too much other things to focus on.