This paper introduces SMPL-X, a new 3D human body model that jointly captures the body, face, and hands. The model extends SMPL by adding fully articulated hands and an expressive face. To estimate the 3D model from a single monocular image, the authors propose SMPLify-X, an approach that improves upon SMPLify by detecting 2D features for the face, hands, and feet, training a new neural network pose prior, defining a new interpenetration penalty, automatically detecting gender, and using a faster PyTorch implementation. The model is evaluated on a new dataset of 100 images with pseudo ground-truth, showing improved accuracy compared to existing methods. The models, code, and data are available for research at https://smpl-x.is.tue.mpg.de. The paper also discusses related work, including previous models for body, face, and hand modeling, and presents a technical approach for fitting SMPL-X to single RGB images. The results show that SMPL-X provides more expressive and realistic 3D reconstructions of bodies, hands, and faces from monocular RGB data.This paper introduces SMPL-X, a new 3D human body model that jointly captures the body, face, and hands. The model extends SMPL by adding fully articulated hands and an expressive face. To estimate the 3D model from a single monocular image, the authors propose SMPLify-X, an approach that improves upon SMPLify by detecting 2D features for the face, hands, and feet, training a new neural network pose prior, defining a new interpenetration penalty, automatically detecting gender, and using a faster PyTorch implementation. The model is evaluated on a new dataset of 100 images with pseudo ground-truth, showing improved accuracy compared to existing methods. The models, code, and data are available for research at https://smpl-x.is.tue.mpg.de. The paper also discusses related work, including previous models for body, face, and hand modeling, and presents a technical approach for fitting SMPL-X to single RGB images. The results show that SMPL-X provides more expressive and realistic 3D reconstructions of bodies, hands, and faces from monocular RGB data.