Ruyi-Zha / endosurf

MIT License
47 stars 3 forks source link

Do we need dynamic modeling for the SCARED dataset? #8

Closed yifliu3 closed 9 months ago

yifliu3 commented 9 months ago

Dear authors,

Thanks for your great work! When reading the paper, I find that SCARED contains static scenes, which seems to follow the typical satatic setting of NeRF. My questions is do we still need dynamic modelling for this dataset and can other static methods work for this dataset?

Ruyi-Zha commented 9 months ago

Thanks for your interest in our work. The SCARED dataset is actually for static stereo matching so there are many other works for this dataset. We disable the deformable part by setting use_deform to False in configuration files, similar to the NeRF setting.