PozzettiAndrea/ComfyUI-MotionCapture

ComfyUI wrapper for Motion capture from video

What it solves

It enables the extraction of 3D human motion and skeletal data from standard video files within the ComfyUI ecosystem. This allows users to capture realistic human movement for use in 3D animation or digital characters.

How it works

The project implements a set of custom nodes for ComfyUI based on the GVHMR (World-Grounded Human Motion Recovery) framework. It leverages state-of-the-art pose estimation tools, including ViTPose and HMR2, to extract SMPL and SMPL-X body model parameters from video input.

Who it’s for

Digital artists, animators, and AI video creators who use ComfyUI and need a way to convert real-world video footage into 3D skeletal motion data.

Highlights

  • 3D human motion capture from video.
  • Based on the GVHMR framework for world-grounded motion recovery.
  • Support for SMPL and SMPL-X body model parameters.
  • Integrated as a custom node package for ComfyUI.

Related

  • Project
    MrGiovanni/AbdomenAtlas

    A large-scale multi-organ CT dataset and toolset for efficient abdominal segmentation, reducing annotation time from decades to weeks using human-in-the-loop AI.

  • Project
    arthurflor23/handwritten-text-recognition

    A TensorFlow-based framework for handwritten text recognition and synthesis, featuring extensive data augmentation and support for multiple standard HTR datasets.

  • Project
    allenai/olmoearth_pretrain

    A family of multi-modal, spatio-temporal foundation models for Earth Observation, providing a scalable framework for planetary intelligence using satellite data.

  • Project
    zae-bayern/elpv-dataset

    A benchmark dataset of 2,624 annotated electroluminescence images of solar cells used for the visual identification of defects that reduce power efficiency.