Hu Wenbo comments

Results 15 comments of


                                            Hu Wenbo

Why the linear discriminator D works

Thanks for your quick response.

Training time

Thanks very much for the kind reply! BTW, is the log for 5cm voxel available, since it seems too long to train a model for 2cm voxels..

Questions regarding to the PSNR/SSIM results

Could you please share your trained model @eedalong . Thanks a lot.

Without the BN layer can performance well

You could put the quantitative results and the trained model for evaluation.

Generate depth map

> Thank you for your work and contribution! I want to generate a depth map, in the "neural_points_volumetric_model.py " file, I set the self.return_depth is True, however, there is a...

Tensorflow Re-Implementation

Hi @MahmoudSelmy . Could you please tell me Where to find the training data. Thank you very much

the result on BSD100

Hi @thstkdgus35, If I put the 'benchmark' folder in src/data/, and I want to generate the benchmark results reported in the paper, I run 'python main.py --data_test Set5 --scale 4...

eval_img error

It seems the test ray_chunk is all sampled at empty space, did you use the default hyper-parameters? And what's your GPU type, if I can reproduce this bug I can...

About the use of custom data resulting in insufficient CPU memory and poor performance

Q: how can I reduce CPU mem? A: The current implementation caches all rays and pixels in CPU memory for efficiency consideration. You may modify it if your machine doesn't...

How to understand "To sample the cone, we cannot follow MipNeRF [3] to use the multivariate Gaussian, since the multivariate Gaussian is anisotropic but the pre-filtering in our Tri-Mip encoding is isotropic."

'isotropic' means that the mipmap is pre-filtered with a square kernel, which cannot support sample anisotropic areas