V2.0 - Retrained with time shifts for the two models, which seems to clean up some of the artifacts that sometimes popped up. It's a little easier to calibrate the high noise model now, and the motion is more consistent.

If you need starter images I recommend this model

Updated I2V/T2V for Wan 2.2 for paw pose and pet play motions.

I2V: Start with weights High @1-1.5 and Low @ 1.0.

T2V: Weights at 1.0

Seems to work pretty well both with photos in pet poses as well as moving into them by prompting, but definitely helpful to be explicit about the expected motion. With less verbose text it gets a little frantic.