The Parameter Configuration on Tanks and Temples dataset

facebookresearch / localrf

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

MIT License

956 stars 62 forks source link

The Parameter Configuration on Tanks and Temples dataset #47

Closed Xiaohan-Z closed 5 months ago

Xiaohan-Z commented 5 months ago

Thank you for your outstanding work！

May I ask if you can provide me with the parameter configuration for the tank and samples experimental results in the article, such as FOV？

Thank you!

Xiaohan-Z commented 5 months ago

Is the experimental results of tank and samples in the paper averaged results of all scenarios in the dataset? tanks_result

ameuleman commented 5 months ago

Hi, We use 71.08 FoV (corresponding to a focal length of 0.7*width as suggested in the dataset's download page). Results in the paper are averages of nine scenes (see this issue).