LYU

Hi，奎酱，请问在LLFF数据集中你用了NDC空间吗？我注意到NGP_PL生成图片比细节上比没有NDC空间的NERF要好一些，但是PSNR和SSIM不如加了NDC空间的NERF。我在代码里似乎没有找到NDC空间，请问你是怎么做到的呢？另外原始的instant ngp重建出来的LLFF数据集很糟糕，我很好奇NGP_PL是怎么达到和nerf类似的效果？

issue when installing Cuda extension

1

Thanks for the excellent work,but when I try to install cuda extension on windows10 I meet some problem likes :  I found my error is same as #34 ,and...

I cannot install pytorch extension on windows

14

Like a lot of other question,when I try to install pytorch extension on windows i meet many problems,but when i try to find answer on issues,there is no same to...

what does relative loss mean?

1

Hi,I found you add relative_loss in your code. I tried relative_loss,weighted_loss,and nomal mse, then I foud relative_loss is best but I donnot know why. I wonder what does relative loss...

Hi，I used the Tanks and Temple datasets you provide,and I found in Auditorium datasets you delet 4 pictures. I wonder why ?

which field will be better compared with colmap

1

Thanks for this excellent work. But I wonder which field will be better compared with colmap. Will it be faster? or get more accurate pose?