GANimator
GANimator
Overview
Here we provide guidelines for training and evaluating the re-implementation of the GANimator model in moai, as well as for generating the motion features used as input to the model. The original implementation can be found here.
Commands
The documentation is split into 3 sections briefly discussing the configurations behind each of the following commands:
Data Preprocess
Convert the available BVH animation file into motion features:
python -m moai run test /path/to/ganimator/conf/bvh2npz/main.yaml --config-dir ./conf bvh_filename=%input_file_name% export_filename=%choose_a_name%Train GANimator
Train GANimator using the converted motion features:
python -m moai run fit /path/to/ganimator/conf/run/main.yaml --config-dir ./conf npz_filename=%input_npz_file_name%Evaluate GANimator
Evaluate GANimator on the predefined metrics:
python -m moai run test /path/to/ganimator/conf/run/main.yaml --config-dir ./conf npz_filename=%input_npz_file_name% model_ckpt=%path/to/trained/GANimator/checkpoint.ckpt% +mdm_ckpt=%path/to/SinMDM/t2m/text_mot_match/model/finest.tar%Alternativelly, for generating new motions with a trained GANimator without the need for metrics, run the following command:
python -m moai run test /path/to/ganimator/conf/run/main.yaml --config-dir ./conf model_ckpt=%path/to/trained/GANimator/checkpoint.ckpt% +mdm_ckpt=%path/to/SinMDM/t2m/text_mot_match/model/finest.tar% +out_name=%exported_file_name% +export_dir=%path/to/generated/motions%Contents
For more details about each step please select the corresponding card below:
Last updated on