Notes 1: Please input full-body image in case of detection errors. We simplify the pipeline in spaces: 1) using Rembg instead of SAM2; 2) limit the output video length to 10s; For best visual quality, try the inference code on Github instead.
Notes 2: We drop ComfyUI Nodes of LHM on https://github.com/aigc3d/LHM/tree/feat/comfyui which support any character and any driven videos as input. Try it!
Large Animatable Human Model 
