Alignment problems

Write here about construction, ideas, equipment, tips n tricks etc. related to structured light scanning
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: Alignment problems

Post by Micr0 »

spark wrote: 27 Sep 2017, 07:06 i am really exited about the calibration routine of flexscan3d.
The DOF is also beeing considered...

https://www.youtube.com/watch?v=ovnn6sSlSHk
And here's another very similar:

https://www.scaninabox.com/

It too uses a single flat panel and multiple "scans" for calibration. I actually hope to be playing with one later today. I'll report back.

FWIW I did some scanning yesterday with the rig calibrated to 20mm with a fairly shallow camera to projector angle. On this job the subject had to remain stationary and the camera moved around it. I was amazed at just how shallow the sweet spot for scanning was. I don't necessarily think this is a calibration problem but actually is more of projector problem. The projected seems to lose both focus and contrast fairly quickly once you get further or closer from the center of focus. As soon as I get some $ I hope to buy a sacrificial projector and see about adapting a good lens.
µ
3dxcan
Posts: 74
Joined: 02 Jun 2017, 17:21

Re: Alignment problems

Post by 3dxcan »

Micr0 wrote: 28 Sep 2017, 16:50
spark wrote: 27 Sep 2017, 07:06 i am really exited about the calibration routine of flexscan3d.
The DOF is also beeing considered...

https://www.youtube.com/watch?v=ovnn6sSlSHk
And here's another very similar:

https://www.scaninabox.com/

It too uses a single flat panel and multiple "scans" for calibration. I actually hope to be playing with one later today. I'll report back.

FWIW I did some scanning yesterday with the rig calibrated to 20mm with a fairly shallow camera to projector angle. On this job the subject had to remain stationary and the camera moved around it. I was amazed at just how shallow the sweet spot for scanning was. I don't necessarily think this is a calibration problem but actually is more of projector problem. The projected seems to lose both focus and contrast fairly quickly once you get further or closer from the center of focus. As soon as I get some $ I hope to buy a sacrificial projector and see about adapting a good lens.
profesional scanners have projectors with exchangable lens, just like the lens we install on the camera. anyone knows what type of projector is that??
erapip
Posts: 56
Joined: 16 Dec 2016, 07:16

Re: Alignment problems

Post by erapip »

I think that ONE of the most important case for good alignment is also the SETUP. I tested different projection angle and with "Total triangulation angle" 19.18° i had lines in the scan. Recalibrate with "Total triangulation angle" 23.89° and the lines disappear. Also the alignment look better.
I would say that Diagonal Setup gives nice scan details than only Horizontal. Correct me if im wrong!

The new version 5.2.1 is out. Any one tested?

HAPPY SCANNING to ALL
Attachments
2017-09-28_00-52-35.png
2017-09-27_22-39-43.png
3dxcan
Posts: 74
Joined: 02 Jun 2017, 17:21

Re: Alignment problems

Post by 3dxcan »

erapip wrote: 01 Oct 2017, 19:01 I think that ONE of the most important case for good alignment is also the SETUP. I tested different projection angle and with "Total triangulation angle" 19.18° i had lines in the scan. Recalibrate with "Total triangulation angle" 23.89° and the lines disappear. Also the alignment look better.
I would say that Diagonal Setup gives nice scan details than only Horizontal. Correct me if im wrong!

The new version 5.2.1 is out. Any one tested?

HAPPY SCANNING to ALL
thats seems like a nice scan, however you can see your shots are not aligned well as you see single colors on most of the scanned voulme. the scans are not blended together, which shows alignment needs to be better. also in your fused result, borders of each shot is visible. that means scans are misaligned.

Is it a David SLS or a custom setup?
BTW, thanks for notifying about the new release. I'll try to install.
Attachments
2017-09-28_00-52-35.png
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: Alignment problems

Post by Micr0 »

3dxcan wrote: 01 Oct 2017, 19:26
erapip wrote: 01 Oct 2017, 19:01 I think that ONE of the most important case for good alignment is also the SETUP. I tested different projection angle and with "Total triangulation angle" 19.18° i had lines in the scan. Recalibrate with "Total triangulation angle" 23.89° and the lines disappear. Also the alignment look better.
I would say that Diagonal Setup gives nice scan details than only Horizontal. Correct me if im wrong!

The new version 5.2.1 is out. Any one tested?

HAPPY SCANNING to ALL
thats seems like a nice scan, however you can see your shots are not aligned well as you see single colors on most of the scanned voulme. the scans are not blended together, which shows alignment needs to be better. also in your fused result, borders of each shot is visible. that means scans are misaligned.

Is it a David SLS or a custom setup?
BTW, thanks for notifying about the new release. I'll try to install.
I see what you mean, but the parts of the scan that are aligning are aligning quite well. The holes in the bosses look properly round etc. I think if he slightly increased the error rejection and or the outlier filtering that would improve. I have found that the distortion like this in most scans is not linear. So, the center of the scan may actually be relatively accurate. Trim away the sub optimal edges and you can clean up the final result.

What do you mean by diagonal set up? Do you have you cameras above and below the projector but offset to either side too? I have been using a side by side set up mostly for convenience but have previously found that calibration is easier with a vertical camera placement. Again one of the reasons I wish David would rethink the calibration procedure.
µ
3dxcan
Posts: 74
Joined: 02 Jun 2017, 17:21

Re: Alignment problems

Post by 3dxcan »

Micr0 wrote: 02 Oct 2017, 15:10
3dxcan wrote: 01 Oct 2017, 19:26
erapip wrote: 01 Oct 2017, 19:01 I think that ONE of the most important case for good alignment is also the SETUP. I tested different projection angle and with "Total triangulation angle" 19.18° i had lines in the scan. Recalibrate with "Total triangulation angle" 23.89° and the lines disappear. Also the alignment look better.
I would say that Diagonal Setup gives nice scan details than only Horizontal. Correct me if im wrong!

The new version 5.2.1 is out. Any one tested?

HAPPY SCANNING to ALL
thats seems like a nice scan, however you can see your shots are not aligned well as you see single colors on most of the scanned voulme. the scans are not blended together, which shows alignment needs to be better. also in your fused result, borders of each shot is visible. that means scans are misaligned.

Is it a David SLS or a custom setup?
BTW, thanks for notifying about the new release. I'll try to install.
I see what you mean, but the parts of the scan that are aligning are aligning quite well. The holes in the bosses look properly round etc. I think if he slightly increased the error rejection and or the outlier filtering that would improve. I have found that the distortion like this in most scans is not linear. So, the center of the scan may actually be relatively accurate. Trim away the sub optimal edges and you can clean up the final result.

What do you mean by diagonal set up? Do you have you cameras above and below the projector but offset to either side too? I have been using a side by side set up mostly for convenience but have previously found that calibration is easier with a vertical camera placement. Again one of the reasons I wish David would rethink the calibration procedure.
It really depends what you compare the results with. If you compare david with david, yes its pretty acceptable. But if you compare it to professional scanners, the alignment Is off.
You dont see such problems highlighted above in their results.
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: Alignment problems

Post by Micr0 »

3dxcan wrote: 02 Oct 2017, 17:31
Micr0 wrote: 02 Oct 2017, 15:10
3dxcan wrote: 01 Oct 2017, 19:26

thats seems like a nice scan, however you can see your shots are not aligned well as you see single colors on most of the scanned voulme. the scans are not blended together, which shows alignment needs to be better. also in your fused result, borders of each shot is visible. that means scans are misaligned.

Is it a David SLS or a custom setup?
BTW, thanks for notifying about the new release. I'll try to install.
I see what you mean, but the parts of the scan that are aligning are aligning quite well. The holes in the bosses look properly round etc. I think if he slightly increased the error rejection and or the outlier filtering that would improve. I have found that the distortion like this in most scans is not linear. So, the center of the scan may actually be relatively accurate. Trim away the sub optimal edges and you can clean up the final result.

What do you mean by diagonal set up? Do you have you cameras above and below the projector but offset to either side too? I have been using a side by side set up mostly for convenience but have previously found that calibration is easier with a vertical camera placement. Again one of the reasons I wish David would rethink the calibration procedure.
It really depends what you compare the results with. If you compare david with david, yes its pretty acceptable. But if you compare it to professional scanners, the alignment Is off.
You dont see such problems highlighted above in their results.
It is about comparing the final mesh to the real world subject. I have done this by comparing the measurements from the scan to CMM measurement data and found my scans in David to consistently be in the ~.05% range for accuracy. That accuracy seems to decreases exponentially the further you get away from the point at which the camera and projector were focused.
µ
erapip
Posts: 56
Joined: 16 Dec 2016, 07:16

Re: Alignment problems

Post by erapip »

Some data comparison FLEX-->DAVID-->Photogrammetry. Same hardware and same setup.
David for my point of view is the winner (only two scan) :)
The learning curve make the diference.......
Attachments
Untitled.jpg
Default_2.43.jpg
Untitled44.jpg
Untitled3.png
Untitled.png
Untitled.jpg
3.png
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: Alignment problems

Post by Micr0 »

Thanks. Can you post more information/pictures of your set up?

Can you elaborate on why you think David was better?
µ
3dxcan
Posts: 74
Joined: 02 Jun 2017, 17:21

Re: Alignment problems

Post by 3dxcan »

2 shots is too few to judge. Flexscan can utiluze markers for alignment. So instead of millions of points, it aligns using few number of target points. That's where the real benefit is.i'd try with a lsrge object and use target points.
Post Reply