How many images did you used to generate point cloud for each scene?

dunbar12138 / DSNeRF

Code release for DS-NeRF (Depth-supervised Neural Radiance Fields)

https://www.cs.cmu.edu/~dsnerf/

MIT License

749 stars 127 forks source link

Closed rayleizhu closed 3 years ago

rayleizhu commented 3 years ago

e.g. when you use two images to train NeRF, did you also use two images to generate point cloud?

dunbar12138 commented 3 years ago

Yes, we run COLMAP on 2 images for point clouds and cameras when training 2-view NeRF / DS-NeRF.