I’m crashing quite a lot at the moment - seems to be something to do with the control points.
Is there an option to roll back the changes or is this a know issue and we can expect another patch soon?
Thank you
Jennifer (crash dumps uploads and about to some that didn’t make it to the debug@ address
Hi Jennifer Cross
There is no roll back, and even new versions cant solve issues with CP-GCP points if they are WRONGLY placed what is the common cause of the issues. Can take a look with TeamViewer on the data ?
Looks like part of the problem is the CP’s mode allows you to place points outside the image boundary.
I assumed since it allowed it, you must be doing some fancy extrapolation and still provided some location information for the image. Or what it does it make it crash… one of these things takes less coding than the other 
Suggest two features…
one - don’t let the tagging happen outside the image boundary if you aren’t prepared to cope with that.
two - if you encounter loaded/saved points beyond the specified image edges, either discard or disable that point with a warning.
(colour ! triangle again on the image in the control point list?)
I’ll parse the control point file and see how it goes when I clip for boundary manually…
but hey - using position information from points outside the image would be a great addition
Thank you
Jennifer
Hi Jennifer Cross
Looks like part of the problem is the CP’s mode allows you to place points outside the image boundary.
And for what reason you place them out of image ? what sort of workflow “magic” is this ? CP-GCPs are intended to be used on IMAGE and not outside of the image space. Even placing them on “borders” (5-10% from outside ) is and can cause issues.
RED exclamation mark is clearly to be seen on particular image with pix error ( under particular CP-GCP )
Wishgranter wrote:
Hi Jennifer Cross
Looks like part of the problem is the CP’s mode allows you to place points outside the image boundary.
And for what reason you place them out of image ? what sort of workflow “magic” is this ? CP-GCPs are intended to be used on IMAGE and not outside of the image space. Even placing them on “borders” (5-10% from outside ) is and can cause issues.
RED exclamation mark is clearly to be seen on particular image with pix error ( under particular CP-GCP )
Ah - so that is what the red mark indicates - it seemed to be an out of alignment error/point adjustment warning.
As for magic… giving the software a direction vector (out of image) would still provide a good hint for alignment (you couldn’t SLAM with it, but at least you could reduce the search space for features since you could know the nearby features which might be visible. At worst, I’d assume the software would ignore a point if out of bounds/unusable.
If there is a 5+% non-CP-able zone on each picture, maybe display a safe tagging overlay while in CP mode? Should I record some of these over in the suggestions area?
Thank you for the reply - I’ll re-filter the control point file for outliers and give it another go.
Jennifer
Hi Jennifer Cross
Ah - so that is what the red mark indicates - it seemed to be an out of alignment error/point adjustment warning.
As for magic… giving the software a direction vector (out of image) would still provide a good hint for alignment
Would not as this is photogrammetry based algos, working with pixels directly, how you can know its properly placed outside of the image ?? based on what data or just assumption, what sort of ?? Properly placed CP-GCPs work flawlessly. Its just about accepting thsi is photogramemtry based solution and its rules.
If there is a 5+% non-CP-able zone on each picture, maybe display a safe tagging overlay while in CP mode? Should I record some of these over in the suggestions area?
No need, in principle going outside of central part you get worse and worse pixel quality ( Vignetting, distortion, and etc. ) so you try use not the optimal pixels, so its just logical to use the best part of image. here its hard to say what % can be used as its depend on camera + lens + postprocessing of the data and etc. its just logical not ot use border areas of image.
Please take a look on this example of CP-GCP workflow, how are the points spread over the image and etc.
https://www.youtube.com/watch?v=d8naLEtLqDY
Well that is interesting…
I’ve pruned the list of all points outside the image extents.
Re-loaded my 21 points from the csv (made a new project, added the folder of images then added csv of control points)
And it crashes.
try aligning just the pictures, no crashing but get 9 components.
load the CP and they have of the triangle warnings (I assume because the conflict with the alignment?). Delete all image references in the CP lists that are marked with red triangle. Align… crashes again
Fragile… Any suggestions? There are still some gray triangle marked CP (yellow triangles inside)
Thank you
Jennifer
Hi Jennifer Cross
Can take a look with TeamViewer directly on the dataset ? or its possible to send the whole project data including RCROJ files and etc ?