Clearly this is a case of inadequately overlapping or overly converged photography, but I’d like to share varying results I got aligning the same batch of images in RC and in Autodesk Remake. No, this isn’t a shoot out, I’m not returning to Remake, which in other comparisons came in much weaker on resolution, not to mention other limitations on the reconstruction side. Rather, I’m trying to understand why Remake was able to align most of these cameras and RC returned a couple dozen components, none of them containing much that was usable. (I’ve attached a screen grab in Remake to show alignment is possible and looks quite okay).
It’s my understanding a photogrammetry engine is tuned variously to how strict the rules are on residuals for accepting candidate features supporting accuracy of camera poses. Might this be simply what explains the different results? Is there no way to relax that or another variable to allow RC to accept a similar set of aligned cameras as what Remake was able to solve? Such a pref in alignment settings might spell the difference in coming home with workable data versus a forced reshoot.
Thanks,
Benjy