search

EPFL-VILAB / omnidata

A Scalable Pipeline for Making Steerable Multi-Task Mid-Level Vision Datasets from 3D Scans [ICCV 2021]
Other
395 stars 49 forks source link

Depth output #39

Open stalkerrush opened 1 year ago

stalkerrush commented 1 year ago

Hi,

May I know whether the depth in the demo correspond to the z value in camera coordinates or the length of the ray hitting the surface? I also see you distinguish between z-buffer depth and Euclidean depth, would you mind elaborate a bit on the difference between them?

Thanks!

jingyibo123 commented 1 year ago

Euclidean depth is the ray distance from point to camera center z-buffer depth is the point to plane distance from point to camera Z-plane I think in the SLAM community the former is often called 'distance', the latter 'depth'.