Alignment problems

Write here about construction, ideas, equipment, tips n tricks etc. related to structured light scanning
3dxcan
Posts: 74
Joined: 02 Jun 2017, 17:21

Re: Alignment problems

Post by 3dxcan »

Micr0 wrote: 09 Oct 2017, 14:19
Stive_Rad0 wrote: 08 Oct 2017, 18:20
Micr0 wrote: 08 Oct 2017, 13:49 Generally I can tell how good my alignment is by how well David displays the checkerboard after the calibration sequence. This is important feedback. The key is also looking at the alignment of the corners that are at the very periphery of the envelop. If they aren't perfect, Adjust the position of the equipment to the corner and try again. This may mean a slightly wider camera/projector angle etc. If your calibration corners aren't perfectly made there are parameters in the advanced settings to help compensate for this but, I stay away from those unless absolutely necessary.
Interesting. for me, some of the projected corners are over 1mm deviated from circle centers. I'm sure my calibration board is made perfectly, but over 1mm error is unacceptable. it is interesting to say with this error I can get alignment well. BUT only with dual camera and adjusting cameras exactly at the same angle. (with single camera corner deviation is the same, but alignment error is over 2mm). Why I can't get Corners at center of circles?
Projector problem or something else?
Are your advance settings all set to the default?

P.S. The first set of panels I make I thought were perfect too. Then I found a simple mistake in the parts that hold the panels that was throwing the whole thing off.
For me, the calibration checkered pattern is jagged and can't be jugged if it crosses the center of calibratiom circles. Using LG full hd tho, its smooth. Maybe its a good idea to print the center point in different color to get better idea of the calib precision.
Stive_Rad0
Posts: 35
Joined: 17 Jun 2017, 18:04

Re: Alignment problems

Post by Stive_Rad0 »

3dxcan wrote: 09 Oct 2017, 17:23 For me, the calibration checkered pattern is jagged and can't be jugged if it crosses the center of calibratiom circles. Using LG full hd tho, its smooth. Maybe its a good idea to print the center point in different color to get better idea of the calib precision.
If your Projected Check Board is Jagged, it is a projector problem.
3dxcan
Posts: 74
Joined: 02 Jun 2017, 17:21

Re: Alignment problems

Post by 3dxcan »

Stive_Rad0 wrote: 09 Oct 2017, 19:06
3dxcan wrote: 09 Oct 2017, 17:23 For me, the calibration checkered pattern is jagged and can't be jugged if it crosses the center of calibratiom circles. Using LG full hd tho, its smooth. Maybe its a good idea to print the center point in different color to get better idea of the calib precision.
If your Projected Check Board is Jagged, it is a projector problem.
whats your projector resolution?

Acer P1283i native resolution is 1024x768, regular lines look fine , only David Checker board projection is jagged.(see aliased image http://www.game-debate.com/blog/images/ ... 403272.png)
maybe the checkered pattern used by david isn't optimized for this resolution. however thats used for evaluation of the calibration only and the SL patterns themselves are fine.
Stive_Rad0
Posts: 35
Joined: 17 Jun 2017, 18:04

Re: Alignment problems

Post by Stive_Rad0 »

Hi. Is there any New Solution. I've just tested a new Projector. but with no improvement.
most errors occur in depth, not in a narrow range of depth or a plane zone.
Stive_Rad0
Posts: 35
Joined: 17 Jun 2017, 18:04

Re: Alignment problems

Post by Stive_Rad0 »

Hi again. dose any one find any new solution to reduce distortion problem?
bigbomber
Posts: 114
Joined: 14 Dec 2017, 18:08

Re: Alignment problems

Post by bigbomber »

Hi to all. I had alignment problems for years, but now after a lot of researches and experiences I've got ride of problem. For who are interested in a conclusion of my experiences and solutions from new and old David/HP forums:

two things are the major sources of alignment problems:
1-Calib Panel
2-Projector

How to distinct the problem source:
If your single scans quality is good, Check board edges (projected after calibration process) are not jagged even with one pixel, check board vertexes projected exactly at the center of circles, so you can think about accuracy of Calib panel. 80% problem is coming from Calib panel.
Maybe your calib panel is good, but you need to adjust contrast and brightness of cameras to see & read Calib panel correctly.

If you experience very low jagged edges (max one pixel on a few of edges) on check board (projected after calibration process), the source of problem can be Calib panel or not compatible Projector. for Calib Panel, this error occur only when circles are not set in their correct position on Calib Panel.

If the circles of Calib Panels set correctly at position and you don't have projector problem, even if 90deg of Calib Panel is deviated for some degrees, and Left to Right Panel has other dependence deviations (Offset &/ rotation), software can exactly compensate for it well. But if there is one of the last problem, by activation of Compensation options, you may get very strange and unrepeatable results.

Both Negative and Positive Calib Panel works well. I prefer to use negated panel for my scans.

For dual Camera setup, you do not need to use similar cameras. you can use two separate cams with different resolution, frame rate and lenses.
sometimes I use 8mm lens for one of my cam and 12mm for the other one, and by switching Extended-Mode, I can get deep & wide Scan at the same time.

For distinction of projector problem, the most visible sign is jagged edges on Check Boared and vertexes of cubes not align to center of circles (note it also can come from not correctly positioned circles of Calib panel). if the projector problem was not so acute, you may not attend to jagged edges, but may experience wavy scans and alignment problems which never can solve it by changing setting and Calib panel.
Sometime projector problems prevent software to detect of Negative/Positive corner automatically if you checked AutoDetectCalibCornerSetup and using Negatve corner. Also sometimes and error of "Calibration Panel is not made accurately" shown during calibration process.
Noisy Red Sine Lines can be a result of projector problem.
You must turn off all of auto functions of projector and set all values to neutral and 0.
Some projector have option of Phase and Clock in their menu, which automatically get values when you connect projector to PC. So first plug projector to PC and then check if their value is 0.
Keystone must be set in 0.
projector is a complicated part of our scanners, unfortunately some of them never can be adjusted for 3d scanning process due to their internal process flow and low-level settings, and they need adjustments on internal firmware (as mine).
I think laser scanner can reach to higher accuracy than SLS system, but SLS is more popular and faster.

I will keep to update this post to complete it...
Stive_Rad0
Posts: 35
Joined: 17 Jun 2017, 18:04

Re: Alignment problems

Post by Stive_Rad0 »

Does any one try other solution on alignment problems? I noticed in some situations the problem can be reduced but I can't find any rule for it.
I've also try some other projector from NEC and Optoma, but never can get ride of mesh distortion.
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: Alignment problems

Post by Micr0 »

Stive_Rad0 wrote: 06 Mar 2018, 22:16 Does any one try other solution on alignment problems? I noticed in some situations the problem can be reduced but I can't find any rule for it.
I've also try some other projector from NEC and Optoma, but never can get ride of mesh distortion.
I don't recall, are you using cal panels from David or ones you made yourself?
µ
User avatar
Popex
Posts: 6
Joined: 29 Apr 2018, 00:38
Location: Argentina
Contact:

Re: Alignment problems

Post by Popex »

Hello, as a user of Flexscan3D I have experienced, as David's users, alignment problems.

I have tried with Target, but I have not achieved good results.

Researching about it, I found that in CCAT:

https://www.ccat.us/amc/light-scanning- ... gineering/

They solve this problem aligned with spheres. To do this they import each of the shots in "Geomagic Design X", and they are automatically aligned with the center of the spheres.

I have tried it with bearing spheres painted white (what this adds is an error, as it is not spheres of a white material) and I have achieved good results. Aligned of +/- 0.1mm in a piece of 600mm long. You can see this in the attached image. Green color is within +/- 0.1mm. Yellow between +/- 0.1 and 0.2mm

This I think can be improved, since I lack a sphere on the top of the piece, to improve the alignment. Also being able to work with larger spheres should improve the alignment.

Here the CCAT videos aligned with spheres:

https://www.youtube.com/watch?v=qNfMTatYD2Y

https://www.youtube.com/watch?v=uOPjr59DUmI



This can also be used with David, since the alignment occurs in Geomagic. You use David, Flexscan3D or whatever software, to capture. Then provide the maya in geomagic.
Attachments
Analisis.jpg
Flexscan3D
vondrej
Posts: 1
Joined: 04 Jul 2018, 07:48

Re: Alignment problems

Post by vondrej »

Good afternoon

I am doing research in a company for digitalization stocked parts. My job is decide which method is the best for this purpose (CAD modelling / 3D scanning / photographing). Company has borrowed 3D scan for 2 weeks so i started with it.

I have 3D scan DAVID SLS-2. It is equipped with DAVID cam 3.1 and Acer K132. Software is DAVID3. 1 calibrating panel is broken so I scanned it, printed it and sticked it to carton. To ensure 90 degree iam using that L holders (I don´t know how accurate that printed panel is). In first week i watched lot of videotutorials and read lot of instructions and it looks really simple. By the time I have learned how to configure projector, camera and make successful calibration (I don´t know how can I verify if it is accurate). But i can´t align scans. If I rotate scanned part and make another scan, it looks like that part has different dimensions and align isn´t nice.

I am not sure if i can say if this method is suitable for this purpose in 2 weeks. So Ill be glad if you can help me with anigning and if you tell me your opinion to this situation. If you need i can post how my scans look.

Thank You
Post Reply