Page 1 of 2

alignment problem

Posted: 11 Mar 2017, 21:11
by spark
Hello 3dscanforum,
i hope that you can help me with some trouble that i have with my sls setup.

My setup:
David K135+ projector
IS 23ux174 HD camera 1" sensor
fujinon 16mm megapixel lens
David5
David glass calibration panel
manfrotto tripod and fixtures
turntable
board game figures as sphere targets // target stickers
developers spray (crack detection equippment) in some cases

I tried most of my scans on 120 and 240mm calibration panel.
The scans are okay, could be a bit more detailed, but i need the shape for reverse engineering.
The problem i have are the problems during alignment. Auto alignment doesnt work good, manual alignment is better but still with deviation.
Using geomagic design x the sphere registration is not stable, auto align also fails, manual alignment has the same deviation.
Alignment is not a issue of the software used, its a issue of the scan quality.

For calibration i exactely orientated the panels to 90deg.
Distances are acc. the table stated in sls1 manual, which is easy trigonometry.
Angle is around 20deg.

What i realized is the issue, that my cameralens is too wide and i capture more than the pattern.
I could reach more details with another lens, but is this the problem for alignment?

The components should be capable for good scans, would be nice if you help me with the correct use of them.

Some pictures will follow.
regards, Spark

Re: alignment problem

Posted: 11 Mar 2017, 21:47
by spark
Image
Image
Image
Image
Image
Image
//i can also provide a sample file, if this helps to identify a problem?
regards, spark

Re: alignment problem

Posted: 12 Mar 2017, 14:22
by Micr0
spark wrote: 11 Mar 2017, 21:11 Hello 3dscanforum,
i hope that you can help me with some trouble that i have with my sls setup.

My setup:
David K135+ projector
IS 23ux174 HD camera 1" sensor
fujinon 16mm megapixel lens
David5
David glass calibration panel
manfrotto tripod and fixtures
turntable
board game figures as sphere targets // target stickers
developers spray (crack detection equippment) in some cases

I tried most of my scans on 120 and 240mm calibration panel.
The scans are okay, could be a bit more detailed, but i need the shape for reverse engineering.
The problem i have are the problems during alignment. Auto alignment doesnt work good, manual alignment is better but still with deviation.
Using geomagic design x the sphere registration is not stable, auto align also fails, manual alignment has the same deviation.
Alignment is not a issue of the software used, its a issue of the scan quality.

For calibration i exactely orientated the panels to 90deg.
Distances are acc. the table stated in sls1 manual, which is easy trigonometry.
Angle is around 20deg.

What i realized is the issue, that my cameralens is too wide and i capture more than the pattern.
I could reach more details with another lens, but is this the problem for alignment?

The components should be capable for good scans, would be nice if you help me with the correct use of them.

Some pictures will follow.
regards, Spark
The focal length of the camera length shouldn't matter to the calibration. A couple of observations about the pictures you posted.
first is that it seems the exposure level on the scan subject is a bit low. When setting up, before calibration I set my exposure to the part then adjust up or down the brightness of the projector to not overexpose the calibration panels.

It also looks like you are trying to scan a reflective or slightly shiny surface. is this correct? have you tried coating it? I find that even with David 5 reflective surfaces cause problems. Though not really the ones you seem to be having.

Just to double check, Optimize dual camera calibration is off, yes?

Try calibrating with a less acute vertical angle relative to the calibration panels. That is, try placing the projector and cameras more straight on to the panels.

The surface alignment in the center of your scan field looks fairly good. It looks like the surface curvature is very close to matching but dimensionaly it's off.

Are those panels and panel holder the ones sold by David (OEM)?

Re: alignment problem

Posted: 12 Mar 2017, 14:24
by Micr0
BTW: How big is the part and what size/scale calibration panels are you using?

Re: alignment problem

Posted: 12 Mar 2017, 16:11
by avogra
I think the topic fits better in "Troubleshooting" and moved it there.

Re: alignment problem

Posted: 12 Mar 2017, 20:06
by spark
Hi µ,
your questions:
first is that it seems the exposure level on the scan subject is a bit low.
yes, i am running on max projector brightness. and aperture of 16. the exposure is 1/60s in order to avoid waves.
Amplification is:-4
Image

what do you mean with
When setting up, before calibration I set my exposure to the part then adjust up or down the brightness of the projector to not overexpose the calibration panels.
are you changing the exposure time? /as the multiple from 1/60s? / or are you using the gain setting?
It also looks like you are trying to scan a reflective or slightly shiny surface. is this correct?
yes, it is uncaoated aluminium but low reflective. i tried coating spray on other parts, the acquisition is better but the alignment was similar bad.
I will use coating spray on all future tryouts...
ust to double check, Optimize dual camera calibration is off, yes?
japp!
Try calibrating with a less acute vertical angle relative to the calibration panels. That is, try placing the projector and cameras more straight on to the panels.
next time i will take care about this!
Are those panels and panel holder the ones sold by David (OEM)?
yes original ones.
How big is the part and what size/scale calibration panels are you using?
approx 400mm long and 240 pattern

//by the way: what is the reason for tilting the projector relative to camera? i implemented it like in the sls2/3. angle by instinct.

regards, spark

Re: alignment problem

Posted: 12 Mar 2017, 20:16
by spark
what is the reason for inaccurate acqusition of target stickers?
partly good partly bad...
Image
Image

outliner removal is set to 0.

regards, spark

Re: alignment problem

Posted: 12 Mar 2017, 23:56
by VDX
... it's mostly the colour/absorption or reflectivity -- pure black and shiny/mirroring surfaces are hard to scan correct,

Viktor

Re: alignment problem

Posted: 13 Mar 2017, 06:39
by SilverGhost
pictures of david program shows the resolution of your acer k135+ as 1280x1024 while the native is 1280x800...

Re: alignment problem

Posted: 13 Mar 2017, 11:21
by spark
The targets are special 3d scanning targets and should be recognized fairly good?!

sorry, my projector is the 132+ but the point with the resolution is the same! :oops:
it was choosed by the system as default and i havent questioned about it!
Thank you! i will shortly reset this!

regards, spark