facebookresearch / banmo

BANMo Building Animatable 3D Neural Models from Many Casual Videos
Other
539 stars 58 forks source link

Extracting root pose #71

Closed samiragarwala closed 8 months ago

samiragarwala commented 1 year ago

Hi! Thank you for the great work. I wanted to extract the optimized root poses for one of the cat-pikachiu sequences and was wondering if there was a way that you would recommend to do that? Thank you!

gengshan-y commented 1 year ago

Hi! The easiest way to go is probably using the visualization script for root poses

python scripts/visualize/render_root.py --testdir path-to-the-log-dir

This script loads the optimized root poses at each checkpoint and visualizes them. You may find the vars corresponding to the latest checkpoint, then vars["rtk"]in this line is a Tx4x4 matrix where the Tx3x4 component is the [R|T] matrix defined as object-to-camera transformation.