Page 1 of 2

360° alignment issue and calibration settings questions

Posted: 09 Apr 2020, 19:58
by mr_cg
Hi all,
i am struggling with alignment issues since quite a long time. I getting almost frustrated and i am hoping somebody here in the forum can help me.
I have original HP SLS Pro S3 setup with original HP high quality calibration panels and HP turntable.
The main problem i have is to align 360° scans where first and last scan will be merged. I have already checked all alignments of the scans but here i do not see any major issues ... only at the end.

Long time i have scanned by using all advance settings on default values.
Later on i have activated the option "CompensateAngelBetweenPatterns" and the situation improved but it's still not on acceptable level.
Settings.JPG
I was quite surprised to see that even with original HP scan panels the angle was way off.
From calibration log it shows 90.2367 degree. I guess the situation improved because i have activated the compensation.

Code: Select all

Generate geometry... done.
Generate geometry... done.
>>>>>>>>>>>>>>>>>>>>>>>>
Generate geometry... done.
Generate geometry... done.
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Generate geometry... done.
Generate geometry... done.
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Trying to connect forcing that format...
Connected.
Trying to connect forcing that format...
Connected.
     -> Setup is 'inner corner'
 - RMS error is ok: 0.226991
Camera 0 [HP 3D 2.3MP Cam Pro (15710621)] calibrated:
<camera_model>CameraModelTsai</camera_model>
<cx>1010.2246168842227689</cx>
<cy>585.53833607535648298</cy>
<f>4354.1856709913208761</f>
<sx>1</sx>
<kappa1>4.4759564916411224033e-09</kappa1>
<resX>1920</resX>
<resY>1200</resY>
<nx>0.88125862805627197538</nx>
<ny>0.10050500438179138518</ny>
<nz>-0.46182461451355499849</nz>
<ox>0.19958382592239376496</ox>
<oy>-0.96486814595253311833</oy>
<oz>0.17086766035824654808</oz>
<ax>-0.42842680460792503805</ax>
<ay>-0.24275132341624336418</ay>
<az>-0.87035755185619290142</az>
<px>173.0579443546631353</px>
<py>82.228848598560148275</py>
<pz>320.19647616190292183</pz>

Horizontal open angle: 24.9178
Vertical open angle: 15.721
Calibrate Projector.
~
~
~
CalibrateProjector: leftBackPlane  = (p=-1.12035e-17 -1.12035e-17 22.317 1 n=1 -0 -6.12323e-17 0)
CalibrateProjector: rightBackPlane = (p=22.1219 -0.146048 -4.63094e-17 1 n=0.00413175 2.52997e-19 0.999991 0)
Angle between calibration panels: 90.2367 degree
  Size of camera's center pixel at background plane: 0.123693 mm
cy=-0.5 rmsError=0.00334046
cy=0    rmsError=0.00334048
cy=0.5  rmsError=0.000527335
cy=1    rmsError=0.000527335
cy=1.5  rmsError=0.000527335
  172 outliers removed (total 10042)
  262 outliers removed (total 10042)
~
~
~
<camera_model>CameraModelTsai</camera_model>
<cx>0.51304923588548601021</cx>
<cy>0.94457566639232448669</cy>
<f>2.48652325779162231</f>
<sx>0.62587264757847482777</sx>
<kappa1>0.0058676394785032752119</kappa1>
<resX>1</resX>
<resY>1</resY>
<nx>0.6231397100524080912</nx>
<ny>-0.0087937903437710907006</ny>
<nz>-0.7820611043947849339</nz>
<ox>0.33086598750315715511</ox>
<oy>-0.90308857628412897434</oy>
<oz>0.27378590485754655814</oz>
<ax>-0.70867806518147680173</ax>
<ay>-0.42936428896275796729</ay>
<az>-0.55984078745134657762</az>
<px>238.10725495334736479</px>
<py>70.026041867804195817</py>
<pz>186.36530639315995472</pz>

Saving C:\Program Files\DAVID 3D SCANNER 5\Settings\UserSettings.xml ...done.
Scan of calibration pattern with Quality Check = 1:
CalibScan_QC_1.0.JPG
CalibScan_QC_1.0_2.JPG
Scan of calibration pattern with Quality Check = 0.8:
CalibScan_QC_0.8.JPG
CalibScan_QC_0.8_2.JPG
Scan of calibration pattern with Quality Check = 0.6:
CalibScan_QC_0.6.JPG
CalibScan_QC_0.6_2.JPG
To get smooth and closed surface i need to scan with Quality Check 0.6-0.8.
Scanning with Quality Check = 1 is almost not possible because amout of points not enough.

The scanned object in this example is about 100mm long and i have used the 60mm pattern.
Scan_Example.jpg
I have already tried some different settings in the advanced settings menu but i could not further improve the situation.
Any idea about which settings could improve the situation?
Is there something more which can be concluded from above calibartion log?

Many thanks for your help!!!
Christian

Re: 360° alignment issue and calibration settings questions

Posted: 10 Apr 2020, 20:02
by mr_cg
I think the HP calibration panal are not aligned perfectly.
Any possibility to compensate this in the software?
HP_Panel.JPG

Re: 360° alignment issue and calibration settings questions

Posted: 10 Apr 2020, 21:42
by OBNRacerMan
CompensateAngelBetweenPatterns - this is the calculation of the real angle between the table panes (if this item is turned off, the program will consider that it is exactly 90 °)
I think this is corrected by the inclusion of CompensatePatternRotInPlane (but, like any compensation, this is a necessary measure if normal scans cannot come out without it).

Re: 360° alignment issue and calibration settings questions

Posted: 19 Apr 2020, 17:39
by mr_cg
I am somehow disappointed about the original scan pattern because they are not aligned very well. I check the original scan pattern from a fried and also here both sides are not aligned very well. With pure eye it can be seen that they are not aliged very good.

Any idea how to align both sides from the original pattern?

Re: 360° alignment issue and calibration settings questions

Posted: 21 Apr 2020, 15:13
by Micr0
mr_cg wrote:
19 Apr 2020, 17:39
I am somehow disappointed about the original scan pattern because they are not aligned very well. I check the original scan pattern from a fried and also here both sides are not aligned very well. With pure eye it can be seen that they are not aliged very good.

Any idea how to align both sides from the original pattern?
The most important parameter of the calibration panels is the accuracy of the dot spacing (and to some degree dot shape). This is what David uses for calibration. If the panels are not perfectly 90° or one panel is rotated slightly, David can compensate for that. Now, Having said that. It's your master calibration so you would want the panels to be the best they can be. Compensating in software is no substitute of accurate panels. I don't have the OEM David panels, so I can't speak to them, but I have made more panels than I care to remember. It wasn't till I made a Hyper accurate set on my VMC that I realized just how important the positional accuracy of the dots was. From your picture, it looks like it is entirely possible that the printing got distorted. if you have a set of calipers, I'd try to measure as best you can, the dot spacing at various places and see if you can find any distortion.

Re: 360° alignment issue and calibration settings questions

Posted: 01 May 2020, 12:51
by mr_cg
@ Micr0
Thanks for your feedback. I do not think the dot spacing or the shape of the dots is an issue on the OEM panel because it looks pretty good. I have also checked this by measuring. What's an issue in spacing is the distance from the corner to the first dots ... do you think this is causing big issues? Beside this the main issues are due to slight rotation of the panels.

It would be good to understand which accuracy problems can be compensated by which setting and how to manually align the original scan pattern so that distance from the corner to the fist points and all angles are accurate.

Thanks for your help.

Re: 360° alignment issue and calibration settings questions

Posted: 04 May 2020, 20:39
by mr_cg
@ Micro0: One more question. You wrote "It wasn't till I made a Hyper accurate set on my VMC that I realized just how important the positional accuracy of the dots was."
Can you tell me how you machined the panel using a VMC/CNC?
I have also a CNC machine but i have no good idea how to get the black dots into the panel. Do you have some pictures for this?
Many thanks for your answer.
Chris

Re: 360° alignment issue and calibration settings questions

Posted: 05 May 2020, 20:11
by Micr0
mr_cg wrote:
04 May 2020, 20:39
@ Micro0: One more question. You wrote "It wasn't till I made a Hyper accurate set on my VMC that I realized just how important the positional accuracy of the dots was."
Can you tell me how you machined the panel using a VMC/CNC?
I have also a CNC machine but i have no good idea how to get the black dots into the panel. Do you have some pictures for this?
Many thanks for your answer.
Chris
I used the largest pieces of Melamine I count fit into my VMC. This gave me a 350mm panel. I used a 1/4" endmill and plunged it .005" deep. Then I filled in the depressions with flat black paint. The toroids shapes, as well as some alignment marks I milled with a 1/8" (~3mm) endmill.
IMG_20180212_210456.jpg

Re: 360° alignment issue and calibration settings questions

Posted: 05 May 2020, 20:16
by Micr0
I also did a set of glass panels that I put 3 layers of sign vinyl on. The bottom 2 layers were black and the top white. I then plunged the 1/4" endmill just through the white (.003" exposing the black. Getting the vinyl completely smooth was a pain though. My VMC can repeat .0002" accuracy fairly consistently, so this was repetitively easy.

Re: 360° alignment issue and calibration settings questions

Posted: 05 May 2020, 20:23
by mr_cg
Thanks Micro0 for the information. How did you mill the bigger circles with white dot in the middle? Based on the official HP printouts the black dot is /30 from the scale. Is the size of the point important or just only the position?