alvinliu0/SSP-NeRF

How to extract latent code volume by scc?

yangfan0421 opened this issue · 2 comments

yangfan0421 commented

We don't see the latent code f ∈ R88 mentioned in paper 3.2. Is there any other data preprocessing step?

alvinliu0 commented

Hi:

Actually, we modify the sparse convolution network structure of NeuralBody (i.e., modify the input and output channel dimension). Hence the concatenation of output feature channels is summed up as 88-dim.

yangfan0421 commented

It means that I need to train a new SCN which using data like NeuralBody but only using head part for feature extraction.