zhengqili / Neural-Scene-Flow-Fields

PyTorch implementation of paper "Neural Scene Flow Fields for Space-Time View Synthesis of Dynamic Scenes"
MIT License
710 stars 94 forks source link

Question about Least Kinetic Motion Prior #36

Open rliu100 opened 1 year ago

rliu100 commented 1 year ago

Hi, I was wondering why

sf_sm_loss += args.w_sm * compute_sf_lke_loss(ret['raw_pts_ref'], 
                                                    ret['raw_pts_post'], 
                                                    ret['raw_pts_prev'], 
                                                    H, W, focal)

is called twice at the following two lines?:

https://github.com/zhengqili/Neural-Scene-Flow-Fields/blob/d4001759a39b056c95d8bc22da34b10b4fb85afb/nsff_exp/run_nerf.py#L589

https://github.com/zhengqili/Neural-Scene-Flow-Fields/blob/86ad6ddd1ce1c758bc908ef022ce843aae323d50/nsff_exp/run_nerf.py#L593

Should compute_sf_lke_loss compute the $L_{temp}$ term?

Thank you!

Nole21 commented 8 months ago

Have you figured it out? I have the same question.