Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

about the geo_feat_dim #5

Open
RedemptYourself opened this issue Oct 2, 2023 · 4 comments
Open

about the geo_feat_dim #5

RedemptYourself opened this issue Oct 2, 2023 · 4 comments

Comments

@RedemptYourself
Copy link

Hi ,what an excellent work
我简单的将TriMipRF 里的geo_feat_dim和feature_dim提升到了31和32,在训练过程中没有出现问题,但是到达eval_step执行eval_img报错:
image
将geo_feat_dim重新设置为15不会出现该问题,您是否有头绪?万分感谢

@wbhu
Copy link
Owner

wbhu commented Oct 9, 2023

It may be an issue of tinycudann, you may set a smaller value for Trainer.test_chunk_size to try it

@aiyb1314
Copy link

aiyb1314 commented Nov 7, 2023

I'm experiencing a similar situation, but it's caused by a sample point of 0. I'm not sure how you solved it? Any ideas on how to fix it please?

@AI-slam
Copy link

AI-slam commented Apr 25, 2024

Referen

I also meet this problem, have you solved it?

@Luh1124
Copy link

Luh1124 commented Aug 14, 2024

Not a network issue, it's because during testing, the synthetic data is at the center of the scene, causing the initial chunksize ray sampling to be invalid, as they might just be the first few lines of pixels, all of which are blank. You can simply skip these intervals. You can refer to:

nerfstudio-project/nerfacc#222 (comment)
nerfstudio-project/nerfacc#207 (comment)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

5 participants