Reprojection error question

It looks like new posts in old threads to do not come to the top of the forum. So I have a question about https://support.capturingreality.com/hc/en-us/community/posts/115000780131-2500-images-Infinite-process?page=1#community_comment_115000844851

Wishgranter: “the reprojection error does not need to be ultra-low … everything under or slightly over 0,5 pix is OK”

Manuel: “… setting the reprojection error to 2.0 (which equals to 0,5pix?)”

My question: Is that correct - reprojection error is reciprocal of ‘pix’?

Or is the 0.5pix mentioned by Wishgranter actually the mean and median error figures in Alignment Report?

Hi Tom,

yes, the old-posts-not-showing-up-issue is a bit counterproductive - could you post a feature request or bug report on it? There might be one already since I remember someone else pointing it out.

I am pretty certain that Wishgranters 0.5 is, as you sais, the error in the alignment report. But to achieve it it is sometimes necessary to lower the Max repro error threshold.

 

 

Hello Tom,

in the alignment settings Max feature reprojection error is in pixels, it is an internal precision level used during alignment. It means that if you set this value to 2, the Maximal error after the alignment (in the Alignment report) will not exceed 2 pixels. I agree with Götz, I believe Wishgranter meant the Mean error in the alignment report.

Thank you both

To complete the answer, a good approach is to start with a tolerance of 2. Then, if there are enough points, lower it to 1.5 or 1 to increase the precision. From experience, a tolerance of 0.75 gives an average margin of error of 0.20-0.25. but, of course, the number of points decreases dramatically. But starting with a tolerance of 2 avoids creating multiple components.