Inputs :<ul><li>Frames</li><li>Processing resolution</li><li>Seed</li><li>Steps</li><li>CFG</li><li>Model and loras</li><li>Video to imitate</li></ul>You can also tweak :<ul><li>Skip frames to select a specific part of the input video</li></ul><ul><li>K Sampler algorithm and details</li><li>Add postprocessing such as bigger upscaling, face detailer etc.</li><li>Play around with the Control Net weights ; 0 depth and 0 lineart is better for some cases</li><li>Add a motions lora</li></ul>Don't forget to install the custom nodes with ComfyUI manager.

Simple video2video workflow

<ul><li>Basic flow for video2video imitation poses</li></ul>

Files with SHA256: 8559ffe602c8b7274d58250854dfd09b67ce3404c720181c361802fa68c8c64b