Although there is a preset for my camera in LB, today I tried the new lens calibration.
In contrast to an earlier attempt, today there were no problems finding and recognizing the “new” Apriltags. I chose to make 25 different pictures with the test card and chose the calculated calibration with the best values, which I then saved.
3 different models are calculated, but unfortunately there is only the option to save one. For comparison purposes, it would be nice to be able to save and load all 3 as needed/desired.
Then I ran the old familiar alignment procedure, and tried again to hit the targets as well as possible, in full zoom mode.
To evaluate the result, I have different “shoot and compare” tests.
They all involve measuring how precisely the camera captures certain test shots.
Today I used an engraved/filled Ø5mm circle, which is placed with the help of the “Dock function” with 10mm Padding, in each corner and one in the middle of my 600x400 machine bed.
The deviation over the entire area is less than 0.5mm. This is a clear improvement to my already good results!
The pictures show the 5 tests and a small practical application example at the end.
I measure the deviation by adjusting the adjustment buttons for X and Y in 0.1mm steps.
I have chosen not to fine-tune/save the X and Y deviation compensation, because the inaccuracy of my system is not linear. That is, a correction in TL with 0.25mm does not match the deviation at BR or vice versa.
I am convinced that when I get my new LB camera with better hard/firmware (?) which is supposed to handle exposure much better, I will get even better precision.
The new way of calibrating the lens works perfectly for me and is a definite improvement. Many thanks to the developers.
The alignment procedure could be improved a bit or made simpler in hitting the targets that form one’s visible working area for the camera, but on the other hand, it works fine enough, otherwise my result will not be as precise as it is. 