A set of tools to visualize and interact with sequences of 3D data with cross-platform support on Windows, Linux, and Mac OS X.
- Easy to use Python interface.
- Load SMPL[-H | -X] / MANO / FLAME sequences and display them in an interactive viewer.
- Manually editable SMPL sequences.
- Render 3D data on top of images via weak-perspective or OpenCV camera models.
- Built-in extensible GUI (based on Dear ImGui).
- Prebuilt renderable primitives (cylinders, spheres, point clouds, etc).
- Render videos of the currently loaded sequences.
- Headless/Offscreen rendering.
- Support live data feeds and rendering (e.g., webcam).
- Modern OpenGL shader-based rendering pipeline for high performance (via ModernGL / ModernGL Window).
aitv_sample_motion.mp4
Basic Installation:
pip install aitviewer
Or install locally (if you need to extend or modify code)
git clone git@github.com:eth-ait/aitviewer.git
cd aitviewer
pip install -e .
Note that this does not install the GPU-version of PyTorch automatically. If your environment already contains it, you should be good to go, otherwise install it manually.
The viewer loads default configuration parameters from aitvconfig.yaml
. There are two ways how to override these parameters:
- Create a file named
aitvconfig.yaml
and have the environment variableAITVRC
point to it. Alternatively, you can pointAITVRC
to the directory containingaitvconfig.yaml
. - Create a file named
aitvconfig.yaml
in your current working directory, i.e. from where you launch your python program.
Note that the configuration files are loaded in this order, i.e. the config file in your working directory overrides all previous parameters.
The configuration management is using OmegaConf. You will probably want to override the following parameters at your convenience:
datasets.amass
: where AMASS is stored if you want to load AMASS sequences.smplx_models
: where SMPLX models are stored, preprocessed as required by thesmplx
package.export_dir
: where videos and other outputs are stored by default.
View an SMPL template
from aitviewer.renderables.smpl import SMPLSequence
from aitviewer.viewer import Viewer
smpl_template = SMPLSequence.t_pose()
# Display in viewer.
v = Viewer()
v.scene.add(smpl_template)
v.run()
Check out the examples for a few examples how to use the viewer:
load_3DPW.py
: Loads an SMPL sequence from the 3DPW dataset and displays it in the viewer.
load_AMASS.py
: Loads an SMPL sequence from the AMASS dataset and displays it in the viewer.
load_DIP.py
: Loads an SMPL and IMU sequence taken from the TotalCapture dataset as used by DIP.
load_obj.py
: Loads meshes from OBJ files.
load_ROMP.py
: Loads the result of ROMP and overlays it on top of the input image with either a weak-perspective or an OpenCV camera.
load_template.py
: Loads the template meshes of SMPL-H, MANO, and FLAME.
load_VIBE.py
: Loads the result of VIBE and overlays it on top of the input image.
quick_start.py
: The above quickstart example.
render_primitives.py
: Renders a bunch of spheres and lines.
stream.py
: Streams your webcam into the viewer.
vertex_clicking.py
: An example how to subclass the basic Viewer class for custom interaction.
The following projects have used the AITViewer:
- Dong et al., Shape-aware Multi-Person Pose Estimation from Multi-view Images, ICCV 2021
- Kaufmann et al., EM-POSE: 3D Human Pose Estimation from Sparse Electromagnetic Trackers, ICCV 2021
- Vechev et al., Computational Design of Kinesthetic Garments, Eurographics 2021
- Guo et al., Human Performance Capture from Monocular Video in the Wild, 3DV 2021
- Dong and Guo et al., PINA: Learning a Personalized Implicit Neural Avatar from a Single RGB-D Video Sequence, CVPR 2022
If you use this software, please cite it as below.
@software{Kaufmann_Vechev_AITViewer_2022,
author = {Kaufmann, Manuel and Vechev, Velko and Mylonopoulos, Dario},
doi = {10.5281/zenodo.1234},
month = {7},
title = {{AITViewer}},
url = {https://github.com/eth-ait/aitviewer},
year = {2022}
}
This software was developed by Manuel Kaufmann and Velko Vechev and Dario Mylonopoulos. For questions please create an issue. We welcome and encourage module and feature contributions from the community.