This workflow creates a new picture from the input picture then tries to animate it.

Inputs :

• Image to imitate

• Models

• Seeds

• Frames

• Motion bucket id

• and so on...

You can do better upscaling at the outputs and don't be scared to touch the ControlNet weights and IPAdapter model.

You might need first to install the missing nodes and download a img2video model here : https://civarchive.com/models/207992?modelVersionId=234212

These custom nodes for ip2p adapters are used also : https://github.com/cubiq/ComfyUI_IPAdapter_plus