/smplcodec

A simple SMPL library

Primary LanguagePython

SMPL Codec

SMPLCodec is a minimal pure-Python library that provides a standardized way to read and write SMPL(-H/-X) parameters of bodies and animations as .smpl files, and export meshcapde scenes as .mcs files.

See the SMPL Wiki for a general description of the model, and the SMPL-X project page and GitHub for model data and code for the most commonly used version.

Installation

pip install smplcodec

Usage

A .smpl files is simply an NPZ that follows some conventions. It is flat-structured and only contains numeric (specifically int32 and float32) data for maximum interoperability. The library provides a SMPLCodec dataclass which provides convenience methods for reading, writing, and validating SMPL data.

    from smplcodec import SMPLCodec, SMPLVersion, SMPLGender

    # Read a 601-frame sequence from a file
    a = SMPLCodec.from_file("avatar.smpl")

    # The full_pose helper property contains the sequence data
    assert a.full_pose.shape == (601, 55, 3)            
    
    # You can also load AMASS sequences
    a = SMPLCodec.from_amass_npz("motion.npz")

    # Create a new neutral avatar and write it to file
    b = SMPLCodec(smpl_version=SMPLVersion.SMPLX, gender=SMPLGender.NEUTRAL, shape_parameters=np.zeros(10))
    b.write("neutral.smpl")

Meshcapade Scene (MCS) Export

SMPLCodec also provides functionality to export meshcapade scenes as .mcs files (GLTF format with SMPL extensions). The class-based interface makes it easy to create scenes with custom cameras and SMPL bodies:

from smplcodec.mcs import SceneExporter, CameraIntrinsics, CameraPose
from smplcodec.codec import SMPLCodec
import numpy as np

# Create exporter
exporter = SceneExporter()

# Load SMPL data using SMPLCodec
body = SMPLCodec.from_file("avatar.smpl")

# Custom camera setup
camera_intrinsics = CameraIntrinsics(focal_length=1000.0, principal_point=(640.0, 480.0))
camera_pose = CameraPose(
    rotation_matrix=np.eye(3, dtype=np.float32),
    translation=np.array([0.0, 0.0, -5.0], dtype=np.float32)
)

# Export scene
exporter.export_single_frame([body], "scene.mcs", camera_intrinsics, camera_pose)

For detailed documentation, see MCS_INTERFACE_GUIDE.md.

License

This library is released under the MIT license.