GarmentCode Documentation

GarmentCode is a framework for programming parametric sewing patterns. It provides the PyGarment library, which allows developers to design garments using base types like Edge, Panel, Component, and Interface. The project also includes GarmentCodeData, a dataset of 3D made-to-measure garments with sewing patterns.

Key components include:

• PyGarment: The core library for designing sewing patterns using edge factories and various operators.
• Configurator: A GUI/online tool for garment configuration.
• Data Generation: Tools for running data generation (e.g., using warp).

To use different body shapes when generating patterns via the command line, modify the body_to_use variable in test_garmentcode.py.

Available Options:

• 'neutral' (Default: gender-neutral average body shape)
• 'mean_female'
• 'mean_male'
• 'f_smpl'
• 'm_smpl'

Configuration Locations:

• Measurement Values: Defined in ./assets/body_measurements.
• Body Shape Examples: Reference files are located in ./assets/bodies.
• Measurement Descriptions: Detailed descriptions are available in docs/Body Measurements GarmentCode.pdf.

The pattern_fitter.py script allows you to fix specific design parameters while varying only the body shapes. This is useful for studying how a single garment design interacts with different body types.

Note: The selection of body shapes is NOT randomized. The script traverses the bodies in the specified sample one by one in alphabetical order.

Default Body Set: The script defaults to the 5000_body_shapes_and_measures sample (this can be changed inside the script).

# Fit the design from t-shirt.yaml to 100 different body shapes
python pattern_fitter.py .\assets\design_params\t-shirt.yaml --name t-shirt-fit --size 100

To simulate GarmentCode patterns in Autodesk Maya using Qualoth, you must install Maya, Qualoth, and the pygarment package into the Maya Python environment. Note that installing warp is not required because the Maya UI uses Qualoth simulation instead of the warp simulator.

To install pygarment, use the mayapy executable to run pip within your Maya environment.

mayapy -m pip install pygarment

Before running the synthetic data generation pipeline, complete the following steps:

1. Obtain body shapes: Download a dataset of sample body shapes from GarmentCodeData or create your own using GarmentMeasurements.
2. Configure paths: Set up the system.json file as described in the Installation guide.
3. Customize design parameters: Edit assets/design_params/default.yaml to adjust parameter ranges and probabilities.
   • Example: To restrict generation to upper garments only, set design['meta']['bottom']['v'] to null and design['meta']['bottom']['default_prob'] to 1.0.
4. Configure simulation: Prepare a simulation configuration file specifying material properties, geometry resolution, and rendering hyperparameters.

GarmentViewer is a GUI script for Maya that allows you to load and simulate sewing patterns from JSON files.

To launch the GUI:

1. Open Autodesk Maya.
2. Copy the contents of ./gui/maya_garmentviewer.py.
3. Paste the contents into the Maya Python scripting console and execute it.

# Copy contents of ./gui/maya_garmentviewer.py into Maya Python console

You can generate sewing patterns from design and body parameters using the command line script. By default, running the script will generate a pattern based on the current state of assets/design_params/t-shirt.yaml for a neutral body and save the output to the logs folder.

To customize the output, you must modify the parameters directly within the test_garmentcode.py script.

python test_garmentcode.py

If you need to install the library manually or in editable mode, follow these steps. This process ensures all dependencies from setup.cfg are met and allows for the manual integration of the Warp simulator.

1. Create and activate a Python 3.9 environment.
2. Install the package in editable mode.
3. Build and install the specific Warp version for GarmentCode.
4. Add the root repository to your PYTHONPATH.

conda create -n garmentcode python=3.9
conda activate garmentcode
pip install -e .
# <build and install warp for GarmentCode>

To install the core pygarment library and its standard dependencies, use pip:

pip install pygarment

To generate different garment samples, modify the style parameters in the configuration files (e.g., assets/design_params/t-shirt.yaml).

Workflow:

1. Update parameter values (the 'v:' field under a parameter name) within their valid ranges.
2. Run python test_garmentcode.py.
3. Find the resulting sewing patterns in system['output']/t-shirt_<timestamp>/.

Note on Units: Parameter values are expressed in centimeters (cm) for distances, degrees for angles, or as fractions.

The pattern_sampler.py script generates 2D pattern samples from GarmentCode parametric garment programs and a body shape dataset. It produces design samples fitted to both a neutral body shape and a random shape sample.

Default Behavior:

• Uses the 5000_body_shapes_and_measures folder in system['bodies_default_path'] for body samples.
• Uses assets/design_params/default.yaml for design parameter sets, ranges, and probabilities.

To use different body sets or design parameter files, you must update the script directly.

# Generate a dataset of 100 samples
python pattern_sampler.py --name garmentcodedata --size 100

# Generate a dataset with a batch ID for parallel processing
python pattern_sampler.py --name garmentcodedata --size 100 --batch_id 0