Open ddkats opened 1 year ago
OpenMVS also supports camera rigs, pls see the camera format in Interface.h
However I am not sure what do you expect to happen if you have stereo images? What you have in mind they should be used for? Currently the algorithm uses multiple target images for each reference image, not only one, so it automaticaly selects 8+ images, not only the stereo pair.
I have been reading about how OpenMVS works, and it seems that it uses a PatchMatch algorithm to compute the depth maps. So, I would like to know if we use pairs of stereo images (left and right), will the pipeline expect to parse pairs of images and not individual images? If yes, is there any suggested name for the stereo images (left_frame000, right_frame000)? I'm using OpenMVS with COLMAP (there is a rig BA option to utilize stereo images), but I wonder how OpenMVS can consider a stereo pair.