smplx

What it is and what it does

SMPL-X is a PyTorch module that loads and evaluates a parametric 3D human body model jointly trained for the body, face, and hands. It represents human shape and pose using learned blend shapes and linear blend skinning, with 10,475 vertices and 54 joints including the neck, jaw, eyeballs, and fingers. The model is defined by pose parameters (θ), shape parameters (β), and facial expression parameters (ψ).

The package provides loaders for multiple related models: SMPL-X (full body with face and hands), SMPL+H (body with detailed hands), SMPL (body only), and MANO (hands only). It requires manual download of model files from the project website and is designed for research applications in 3D human reconstruction, pose estimation, and body animation. The package depends on numpy and torch.

Use it for:

• Reconstruct 3D human body, face, and hand geometry from pose and shape parameters in computer vision research.
• Generate synthetic 3D human data for training pose estimation or body shape prediction models.
• Visualize and manipulate human pose and facial expressions in 3D graphics or animation pipelines.
• Transfer pose parameters between different body models (SMPL, SMPL+H, SMPL-X) for cross-model compatibility.
• Extract vertex correspondences between SMPL-X and related models for hand and face analysis.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Loads and manipulates SMPL-X, a unified 3D body model with 10,475 vertices and 54 joints that jointly represents the human body, face, and hands using PyTorch.

Yes, but with strong caveats. Install if you need SMPL-X for academic research and can work around its abandonment. The package itself installs cleanly with low friction, has no known vulnerabilities, and sees regular downloads. However, it has received no updates since 2021-05-26, so expect no support for newer PyTorch or Python versions, no bug fixes, and no maintenance. The non-commercial license also restricts use to research only. Suitable only for researchers who can maintain their own fork if needed.

Install

smplx on PyPI

pip install smplx

uv add smplx

poetry add smplx

Installing smplx

Before you install

Low friction installation via pip. The package depends on numpy and torch, both widely available. However, the project has been abandoned since its last release on 2021-05-26, so expect no maintenance, bug fixes, or support for newer PyTorch or Python versions.

License in practice

Licensed for non-commercial scientific research only. Commercial use, redistribution, or any use outside academic research is prohibited. You must read and agree to the license terms before downloading or using the model files.

Quickstart

pip install smplx[all]

import torch
from smplx import SMPLX

model = SMPLX(model_path='path/to/model', gender='neutral')
body_pose = torch.zeros(1, 63)
output = model(betas=torch.zeros(1, 10), body_pose=body_pose)

Model files (.pkl or .npz) must be downloaded separately from https://smpl-x.is.tuebingen.mpg.de after registration; they are not included in the package.

Verify before relying

• Whether the package works with PyTorch versions released after 2021 without modification.
• Current compatibility with Python 3.10+ given the project's abandonment status.
• Whether the model files remain available for download from the official website.
• Exact number of joints and their configuration across different model variants.

Package facts

Evidence: smplx-0.1.28-py3-none-any.whl

Tags

More Artificial Intelligence packages

Further reading