Open SSground opened 3 years ago
The video render step gets the result from MLP instead of pre-compute MPI like what we did on the web demo.
So, it takes longer to render video.
The video render step gets the result from MLP instead of pre-compute MPI like what we did on the web demo.
So, it takes longer to render video.
If I using Render video mode,can I used pre-compute MPI to rise speed.
The video render step gets the result from MLP instead of pre-compute MPI like what we did on the web demo.
So, it takes longer to render video.
Hi, can you share the rendering code via pre-compute MPI, thanks!
I cannot find the code that I loaded the precompute MPI. I might accidentally delete it 😢
by the way, you can use the WebGL version instead.
Or if you really want the python version. the key idea is to replace the MLP with the sampling from precomputed MPI image. You can look into how we use F.grid_sample
to sample on mpi_c. but you have to do this on the alpha, basis, and coefficient too.
Has anyone managed to solve the precomputed MPI loading? I'm looking to use this framework for a student project
hi, After the model training is over, why does it take 3 seconds to render a picture? Graphics card: V100 32gb。