How to train and test using GT poses

facebookresearch / localrf

An algorithm for reconstructing the radiance field of a large-scale scene from a single casually captured video.

MIT License

973 stars 61 forks source link

How to train and test using GT poses #17

Closed jywu511 closed 1 year ago

jywu511 commented 1 year ago

Hi, thanks very much for publishing your fantastic work! Localrf can estimate the camera poses as well as the radiance fields. However, if I have the pose ground truth, and I want to estimate the radiance fields only, how can I do? I look forward to hearing from you at your convenience. Thank you in advance for considering my questions. Best regards

ameuleman commented 1 year ago

Hi, --with_GT_poses 1 --lr_R_init 0 --lr_t_init 0 disables pose optimization and loads camera parameters from ${SCENE_DIR}/transforms.json.

jywu511 commented 1 year ago

It works! Thanks!

ameuleman commented 1 year ago

Hi, I changed the parameter name with_GT_poses to with_preprocessed_poses to avoid confusion. In our dataset, transforms.json files correspond to COLMAP poses.

kalelpark commented 6 months ago

Thank you for great discussion.