Simple question that sparked heavy debates internally:
With a near perfect dataset (meaning plenty of very high res sharp images from all possible angles), is it absolutely mandatory to have at least a few control points to get near perfect alignment? And does RC automatically generate CPs under the hood, almost exactly like a user would?
Hi Florian13,
if you have a well captured dataset, then the control points for alignment are not needed.
But you can need some to get the model scaled, but this doesn’t have the influence to the alignment.
RC uses image features in the alignment process internally.
Thanks for the blazing fast reply!
When RC identifies the same feature on multiple images, does it internally create a CP similarly to any user would?
If the features are found on various images, then they are used as tie points for images’ alignment. It is similar to control points, but not exactly the same.
Got it. Do you agree that, placing control points manually can potentially introduce tiny errors since humans are unable to identify a 0 width/height point in space on several pictures? (again in the context of a near perfect dataset).
It is not necessary, as there are the weight and precision settings for the control points. So, there is some free space, where the point can be moved after placing and alignment.
The other way is using control points as exact, then it can influence the alignment negatively.
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.