(WAN 2.2) Action from behind

Details

Model description

My first video lora, doesn't work great at this point. The objective is to try to improve motion/animation for anime scenes of "action" from behind. I would only use this in combination with the general nsfw lora - it doesn't work great on its own. In combination with that lora I slightly prefer the results.

I generally use it at around .75 strength. I think there may have been an issue with looping in some clip in the dataset as I see it sometimes makes the loop worse. Lots of room to improve here for sure. FYI it also sort of knows how to animate a cutout window.

Images made by this model

No Images Found.