search

zju3dv / ENeRF

SIGGRAPH Asia 2022: Code for "Efficient Neural Radiance Fields for Interactive Free-viewpoint Video"
https://zju3dv.github.io/enerf
Other
413 stars 28 forks source link

win_size exceeds image extent #34

Closed luoshuiyue closed 1 year ago

luoshuiyue commented 1 year ago

image

I want to ask how to solve this problem?

luoshuiyue commented 1 year ago

I ran it again without this problem.

luoshuiyue commented 1 year ago

The problem still occurs.

haotongl commented 1 year ago

Hello! Based on the image you provided, it appears that the height or width of the rendering image is smaller than 7. Can you please tell me what dataset you are using?

luoshuiyue commented 1 year ago

I used dtu dateset. I just followed the instructions provided, ran the command: python train_net.py --cfg_file configs/enerf/dtu_pretrain.yaml.

haotongl commented 1 year ago

Could you try the evaluation command to see if it can run normally? If it cannot run normally, can you print the shape of the input of the ssim function?

haotongl commented 1 year ago

I cannot reproduce your problem. Some possible reason maybe unmatched version of ssim function. Could you please try to use scikit-image==0.19.3?

luoshuiyue commented 1 year ago

Thank you very very much for your timely help! I tried version 0.19.3 and there is no error yet (program is stilling running). It seems to work. Thanks a lot!

luoshuiyue commented 1 year ago

I cannot reproduce your problem. Some possible reason maybe unmatched version of ssim function. Could you please try to use scikit-image==0.19.3? Thank you very much! This is the correct way to solve the problem.