Using pretrained Beingh0.5 for MANO-based hand action prediction and rendering

[Qxyuijk]

Hi, thanks for releasing this awesome project.

I am interested in using the pretrained Beingh0.5 model for human hand action prediction, specifically following the MANO-based pipeline as seen in the original BeingH0 repository.

My goals are:

1. To predict hand actions (MANO parameters) using the Beingh0.5 checkpoint.
2. To render the predicted actions into videos for visualization.

Which specific scripts or inference entry points should I use for this purpose? Any guidance or a brief pointer to the relevant documentation/code snippets would be greatly appreciated.