Quick Start

Quick Start#

This tutorial introduces how to run the XRTailor program.

Download Assets#

Assets is an directory that contains necessary files for running XRTailor, e.g., garment templates, garment configurations, body models. The assets directory should follow the hierarchy below:

- Assets
    - Garment
        - SMPLH                                               # SMPL and SMPLH share the same template
            - Template                                        # draped garment meshes in canonical space
                - upper_top_female.obj
                - lower_long-skirt_female.obj
                - ...
            - Config                                          # garment attributes, e.g., attached vertex indices
                - upper_top_female.json
                - lower_long-skirt_female.json
                - ...
        - SMPLX
        - JQR
    - Body
        - Model                                               # body models
            - smpl
                - SMPL_FEMALE.npz                             # model weights
                - ...
            - smplh
                - ...
            - smplx  
                - ...                                     
        - Template                                            
            - SMPLH_female.obj                                # body mesh in canonical space
            - SMPLH_mask.txt                                  # determines whether primitives are used to construct BVH
            - SMPLX_female.obj
            - SMPLX_mask.txt
            - ...
    - ...

We provide example assets to test the pipeline, which can be downloaded from XRTailor Assets.

Getting SMPL Models#

Note that if you need to use smpl pipeline, you need to manually download the SMPL/SMPL+H/SMPL-X models. See smpl models setup for more details. The downloaded models should be put under Assets/Body/Model.

Download Binary#

Download binary distribution from release page.

Windows#

We tested our engine on Windows10 with CUDA 11.3. To execute the binary, make sure that you have CUDA 11.3-12.0 installed.

The binary distribution contains the following files:

- XRTailor_X.X.X_binary_WIN64
    - XRTailor.exe                      // program entry
    - engine_conf.json                  // engine config example
    - Alembic.dll Imath-3_1.dll ...     // runtime libraries for XRTailor, leave as it is

To run the simulation, you need to configure both engine config and simulation config. In engine_conf.json, replace the following paths:

{
    "LOG_PATH": ${PATH_TO_YOUR_LOG_DIRECTORY},
    ...
    "ASSET_DIRECTORY": ${PATH_TO_ASSET_DIRECTORY},
    ...
}

The simulation config differs when you run different pipelines and solver modes. We provide some simulation config examples, see smpl pipeline example, gltf pipeline example and universal pipeline example for more details.

There are plenty of tunable parameters in engine config and simulation config, see engine config example and simulation config example for more details.

Once the engine config and simulation config are properly configured, start the engine utilizing the following command:

.\XRTailor.exe --simulation_config=".\simulation_conf.json" --engine_config=".\engine_conf.json"

For more example usages:

.\XRTailor.exe --help

Linux#

We tested our engine on Ubuntu 16.04/18.04 with Nvidia driver 515.65.01, CUDA 11.3. To execute the binary, make sure that you have Nvidia driver 515.65+ and CUDA 11.3-12.0 installed.

The binary distribution contains the following files:

- XRTailor_X.X.X_binary_LINUX64
    - XRTailor                                     // program entry
    - engine_conf.json                           // engine config example
    - libAlembic.so.1.8 librt.so.1 ...           // runtime libraries for XRTailor, leave as it is

To run the simulation, you need to configure both engine config and simulation config. In engine_conf.json, replace the following paths:

{
    "LOG_PATH": ${PATH_TO_YOUR_LOG_DIRECTORY},
    ...
    "ASSET_DIRECTORY": ${PATH_TO_ASSET_DIRECTORY},
    ...
}

The simulation config differs when you run different pipelines and solver modes. We provide some simulation config examples, see smpl pipeline example, gltf pipeline example and universal pipeline example for more details.

There are plenty of tunable parameters in engine config and simulation config, see engine config example and simulation config example for more details.

Once the engine config and simulation config are properly configured, start the engine utilizing the following command:

Simulation with Display#

Once the engine config and simulation config are properly configured, you can execute the simulation utilizing the following command:

./XRTailor --simulation_config="./simulation_conf.json" --engine_config="./engine_conf.json"

For more example usages:

./XRTailor --help

Simulation without Display#

Sometimes we need to run simulation on a remote server that has no display. Unfortunately, OpenGL need a window to create graphical context. Therefore, you need to setup remote OpenGL following the instructions in Remote OpenGL Setup without X to create a virtual display.

Once you have remote OpenGL properly configured, execute the simulation using:

env DISPLAY=:0 ./XRTailor --simulation_config="./simulation_conf.json" --engine_config="./engine_conf.json"
Contents