Some details about Shallow 3D NeRF Network

[Tara8573]

In the second block of the Shallow 3D NeRF Network(Fig 2), the input dim is 64, while the dim of latent code from mapping network is 128. I would like to know how to convert latent code from mapping network from 128 to 64 ? ?

[Tara8573]

Thank you so much!!!!

[PeterouZh]

The A stands for "affine layer", which is implemented by an FC layer.