Since the automatic detection of suggested CPs, it happens increasingly often that all of a sudden a CP which has been fine so far shows incredibly huge errors. The screenshot shows a new low - almost all of the CPs in the project have been arranged in a line (top right).Usually, it’s just a couple of points at a time. It can strike at any time, somethimes not for hours, then 5 times in 20 minutes.
It can be remedied by either select an image and click on a CP (which can be quite tedious) or by updating constraints, which resets all with one click.
Are the control points also incorrectly placed on the images? Did RC rearranged all the CPs in a line when you were adjusting other CP? If no, what step did you make when this happened?
I will try to reproduce the problem so that we can find out the reason of this behavior.
Interesting, I do feel the the software readjusts your CP to try to be more accurate as you are placing more sometimes.
Also sometimes like you said I will get a HUGE error for the CP when there is no error. Normally this goes away on its own when I try to do a realignment. Clear/Reset suggested points…
Honestly, I dont remember what caused it and how I got rid of it exactly. But I do remember this slight bug.
Also, why did you start calling your CP’s PP? Ground control? I still am not sure how ground control works…
ah, glad to hear somebody else is experiencing it too!
The slight shift in numbers when adding CPs is normal though, since with every new image RC gets new information and the calculation is adjusted. You can observe those changes in the 3D view. When you start out with 2 images, the CP might be quite off, especially if the two images are close. Adding a 3rd one, preferably at about 90° to the first ones, the accuracy goes up and the CP shifts to the new position.
Where did you get the PP from? My screenshot?
Yes, those are ground control points measured with a total station. In German surveilling it’s called a P ■■■ p unkt. It’s actually really simple, you just tell RC this CP is at exactly those coordinates. Then RC tries to make that fit, although only as much as it can. It will show you the remaining error, which is quite nifty because if it’s too big, it points to either a problem in the model or the surveying. I once discovered a fault in one of my PPs because RC would not shift it where I wanted it to be. Since then, I trust RC in terms of accuracy very much…
we do not offer telephone or video call support. Here’s a great video on control points https://youtu.be/daDlC3rr8H8 . To help me diagnose any issue with your scene I will need to know as much information about it as possible, especially any information about the control points like how many images each control point has been confirmed on.
Here is another model. It as almost perfect, but then, it started to get weired and all following steps got worse. It is impossible for me to get back to a stable model
ps: this small model has 180 images, 49 CPs, with 2 to 28 images per CP
pps: yes, cyarks video is great, I have seen it before. thanks for suggesting it!
are you using CPs because you’ve got more components? Are these components georeferenced?
Also, it is better, when CPs are on minimally 3 images, otherwise 28 in unnecessary high count. What happen, when you delete these measurements with high errors? Are these measurements on bad aligned images?
Do you do new alignment or update after CPs measurement?
Can you describe your steps to get this weird model? (import images, alignment settings, how are you measuring CPs, etc.)
