Reproduce synthetic nerf result

[Yongjae Lee]

Thanks for sharing this nice code.
I am trying to reproducing the result on the synthetic nerf.
I found that there is no predefined configuration for it, so I used the default settings from sing_train.sh file like,

scene='nerf_synthetic/lego'
exp_name='lego'
voxel_size=0.01
update_init_factor=16
gpu=3

./train.sh -d ${scene} -l ${exp_name} --gpu ${gpu} --voxel_size ${voxel_size} --update_init_factor ${update_init_factor}

After 30k iterations I got 32.05 dB for the lego scene, which is 3.64 dB lower than that in your paper. I think the default setting I used is not optimal training options. Could you share your testing options?

[Tao Lu]

Hi, @nfyfamr , for synthetic nerf, we try two strategies. One is to train with voxel_size=0.0001, update_init_factor=8. The other is, first train with voxel_size=0.0, update_init_factor=16, and then use the point_cloud.ply as the initial point to rerun the training process (rename+copy). Both strategies achieve similar quality while the latter enjoys a smaller storage and faster inference speed. So we adopt the latter in the paper, as described in section 4.2 (p6~p7).

[Tao Lu]

The training scripts are released.

[Yongjae Lee]

Thanks for your response!