search

dunbar12138 / DSNeRF

Code release for DS-NeRF (Depth-supervised Neural Radiance Fields)
https://www.cs.cmu.edu/~dsnerf/
MIT License
759 stars 129 forks source link

Performance on widely spaced rotated views #25

Open luorix1 opened 2 years ago

luorix1 commented 2 years ago

I have been researching many NeRF-related projects recently and have found that a common constraint is that the views used for training cannot be too far apart. Most common seems to be the forward-facing view scenario. Hence, I am curious if you did any testing to see whether DSNeRF performs better with inward-facing images (i.e. 5 views spaced at 72 degree intervals) or if it has the same issues.

Thank you for your interesting work.

dunbar12138 commented 2 years ago

Thanks for your interest!

Reconstructing scenes from sparse views with wide baselines is a really interesting problem. Since COLMAP will probably not give high accurate results with too wide baselines, DSNeRF will probably not perform as well as with the current setup, but should still be better than plain NeRF.