I wish I had the VRAM to do this at a higher base resolution. I can only do it at 480p with one controlnet, wish I could do 2. It didn't really capture the hip motion and upscaling has introduced flicker.
Someone with a better rig could probably make great use of video input with AnimateDiff. It's tantalizingly close for me.
https://litter.catbox.moe/2jczbm.webm 