Dear DAVID Community,

As you are probably aware, the entire DAVID company, team and product range has been integrated into HP Inc.
This forum is now read-only. Relevant content has been migrated to HP and merged into HP's Support Forums on November 1, 2016.

To start new discussions for 3D Scanning, please register and post your new topic at the HP Support Forums

[Gallery] SL results

Discussions about latest DAVID developments, beta versions...

Postby hal » Wed May 11, 2011 10:47 pm

Simon wrote:maybe DAVID3 has a problem to calibrate your projector. This happens here too from time to time. It would help us to solve your problems if you send us the camera images of your pattern sequence for projector calibration (including one shot for camera calibration). Maybe we can soon build a version that allows saving of the image sequence automatically.


Hello Simon, Sorry for the delay, but a big hardware problem stopped my PC for days. Now I'm on the road again! :D

As you ask, I've sent to your mail a zipped file with all the grabbed images that you need (thanks to the beta07).
In this image you can see 3 test with a "positive" standard angle. In the 3° test I've removed the white and shiny sheets with markers. The result seems quite good, but there is a lot of "modular" noise.

MatM_D3_SLMoreTests.jpg

Here down a test with Frequencies at 8 and differents Phase shifts. The regular noise is there, in any cases.

MatM_D3_SL_PhaseShifts.jpg

Here below you can see a test with a "negative" corner. The result at PhaseShift 4 is better than with "positive" corner.

MatM_D3_SL_NegativeCorner.jpg

MatM_D3_SL_PhaseShifts_Negative.jpg

As you can see here above, the results are better than with "positive" standard corner. This means that the reflection of the light is a big problem (as the wise Gunter already said).
Please note a strange behaviour: with more than 7 shift phases, appars strange "inverted" vertices (the many dark points on the mesh). Those aren't holes but seems more that there is a problem of their normals vectors. Any thoughts why this happen?

Well, So the main problem now is to find a way to cut off the noise and have a perfect scan! :D
An other important hint is that we need better results with Frequencies at default value (6) and have more linear and flat phase error sine.
Many thanks in advance for any aid that cames.
Mattia


p.s.: what that I can't undestand is why David 3 can't calibrate correctly only with one orientation of the patterns? I've placed the projector exactly above the camera, so the horizontal patterns (vertical lines) is completely unusefull. But if I calibrate only with one orientation, the calibration success but the result is deformed (noise + huge deforamtion of the geometry). Why? In the future we will be able to calibrate with only one orientation? The interesting thing is that: if I calibrate with both orientation, after I can scan correctly with only vertical patterns...
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Re: [Gallery] SL results

Postby Simon » Thu May 12, 2011 9:02 am

@Mattia:
Thank you very much for your report. I'm happy that the calibration works better now.
hal wrote:p.s.: what that I can't undestand is why David 3 can't calibrate correctly only with one orientation of the patterns? I've placed the projector exactly above the camera, so the horizontal patterns (vertical lines) is completely unusefull. But if I calibrate only with one orientation, the calibration success but the result is deformed (noise + huge deforamtion of the geometry). Why? In the future we will be able to calibrate with only one orientation? The interesting thing is that: if I calibrate with both orientation, after I can scan correctly with only vertical patterns...

The answer is simple: the current implementation of the projector calibration assumes that both orientations are available. And I'm surprised that SL scanning works with a single orientation, since we never tested and optimized it. (Future work) :D
But I think it is not a hugh drawback if the calibration alwas needs both orientations. But the user interface needs to be made more confortable: e.g. alway both orientation for calibration and after that DAVID asks if further scanning should be made with one orientation if possible.
User avatar
Simon
DAVID Developer
 
Posts: 423
Joined: Sun Jan 21, 2007 3:40 pm
Location: Braunschweig - Germany

Postby hal » Sat May 14, 2011 1:12 pm

Hello and Thanks to your kindly answer, Simon.

Well, here an other test of scanning, with a negative corner (the dark side of the Moon, or the dark side of the Force ah ah :lol: ). Frequencies = 8, Shifts = 16, PHT (PatternHoldingTime) = 200.

MatM_ArtifactsIfTheWhitIsTooLight.jpg

The projector-calibration check sounds good, isn't it? :wink:
But seems that if the white fringes are too bright, the noise is worse and artifacts appears. Simon and/or Sven: Do you think that can you fix this problem? when the light seems too much the mesh appear completely distorted... Ok, we must adjust our projectors, but not so easy with coloured objects with dark and light surfaces (for example, is impossible to scan stull like this (whitout coating), at the current state of art of D3). I've tried to scan a dark/light surface with the 4DDynamics's PicoScan and it works correctly, with the same projector and envirormental light (quite dark room). Ok, well, PicoScan create more noise on the black surfaces, but it can survey it and don't "burn" the white areas.
SOLVED - This was my fault! I've forgotten to modify the PHT value in relation to the new exposure value, sorry. Obviously at low PHT and high Exposure values, the software can't catch correctly all the patterns, so deformations appears :oops: .

Regards,
Mattia
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Re: [Gallery] SL results

Postby MagWeb » Sat May 14, 2011 1:29 pm

Hi Mattia,

Did you play with the Brightness value in SL adv.sets? There you can dim down the projector...
Also think 16 shifts @freq 8 may be a bit high. Don´t think your projector can put that out correctly.
(I use to use 3 or 4 shifts)

Suggestion: In my SL Grabber Soft for 2.x, I dimmed the projector by using /mixing RGB Chanels. This way one can adjust the light color AND intensity to the needs of the object. I found this rather useful.

Gunter
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm

Re: [Gallery] SL results

Postby MagWeb » Sat May 14, 2011 2:38 pm

MagWeb wrote:Also think 16 shifts @freq 8 may be a bit high. Don´t think your projector can put that out correctly.


Just to point that out:
Frequency 5 gives 8 sin waves
Frequency 6 gives 16 sin waves
Frequency 7 gives 32 sin waves
Frequency 8 gives 64 sin waves

Your optoma projector gives 480 lines. If you use Frequency 8 you´ll get 64 sin waves.
This means you have only 7.5 lines/phase ( 0.5 forces interpolation>>>waves of slightly different size)
You cann´t get 16 equal shifts into 7.5 lines.

If you draw a pattern of equal gray values:

gray1
gray2
white
gray2
gray1
black
aso

you should have 6 lines/sin
or next option:

gray1
gray2
gray3
white
gray3
gray2
gray1
black

you have 8 lines/sin
you can do the same with 10 lines/sin

On my projector I have 768 lines. With frequency 8: 768/64 =12
means:
gray1
gray2
gray3
gray4
gray5
white
gray5
gray4
gray3
gray2
gray1
black
12 values as calculated... all is fine here.

In theorie, vertical resolution devided by the number of sin waves should give an even number to have the same values increasing and decreasing.
With 480 lines you get that at Frequency 6: 480/16=30 - this is the lowest even number you can get on 480.

Maybe S&S have implemented some clever pieces of code to solve this , but that is what I hope to have understood so far.

Gunter
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm

Postby hal » Sat May 14, 2011 2:41 pm

Thanks Gunter,

Ok, I'll try your suggestions in the nexts tests. Thanks.
My suggestion to S&S to fix the problems with brightness/darkness and to optimize the scanning process, is more related to the possible competition with others scanners. If David can manage better these problems, could be more flexible and usefull in a lot of light conditions.

mmm Why do you say that I have 480 lines? my working res is 1024x768.

Regards,
Mattia
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

New Test

Postby hal » Sat May 14, 2011 2:53 pm

Hi,

here the result of a new test, with the White value adjusted.
Geometry is good, but the regular noise is still there :evil:

MatM_D3_ScanningTestQuiteGood.jpg

As you can see, the mesh is good, but not still perfect as I dream.

MatM_MeshNotPerfect.jpg

Mattia
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Re: [Gallery] SL results

Postby MagWeb » Sat May 14, 2011 3:05 pm

hal wrote:mmm Why do you say that I have 480 lines? my working res is 1024x768.


Don´t you use the Optoma?
Sorry, if I got something wrong, but on their side is said it is WVGA (854 x 480)
If that is right: you should set your second monitor to the native resolution (also if it supports interpolated higher res)

Gunter
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm

Postby hal » Sat May 14, 2011 3:10 pm

Well, sincerly I've discarted the Optoma, for this moment.
As I wrote in the other post about projector, the transition of RGB seems too slow and it create terrible noise.

Now I use an other mini projector that cames with PicoScan from 4DDynamics company.
It have 800x600 native resolution, and now I'm working at 1024x768 interpolated.

Mattia
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Nice result

Postby hal » Sat May 14, 2011 5:34 pm

Hello,

here my latest and best scan result. Still little noise, but acceptable for a beta :lol:

The Phase Error (mmm seems my encephalogram... flat! :lol: ):
MatM_PhaseError.jpg
MatM_PhaseError.jpg (31.53 KiB) Viewed 3896 times

The scan result:
MatM_D3_NiceScanResult.jpg

Camera parameters:
MatM_D3_CameraParameters.jpg

SL parameters:
MatM_D3_AdvSet_SLgood.jpg

I admit that now I'm more satisfied, but I think that the precision (comparison between virtual and real scanned corner) and lines-noise must be solved more deeply.

However compliments to Sven and Simon.
Mattia
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Postby hal » Sat May 14, 2011 6:23 pm

And here a quick result of good aligning and fusion.

MatM_ExampleResult.jpg

MatM_ExampleResultFusion.jpg
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Re: [Gallery] SL results

Postby WalterMo » Sat May 14, 2011 7:37 pm

Hi,
During the last hours I have made a lot of tests and finally found these settings for my ACER K11 projector with its native resolution of 858 x600:

Settings graphic card:
800 x 600, 32 Bit, 96 dpi, 60Hz.

Advanced settings Structured Light:
Orientation: Both
Pattern Mode: d+ps
Frequencies: 8 (at '6' and '7' are fine V formed stripes!!)
Brightness: 255
PatternHoldingTime: 250
MinContrast: 6

PhaseShift
Shift: 4
Multiplier: 3


The camera was the USB color cam from the DAVID shop. To avoid mutually illuminations of the panels I had put a black cardboard into the corner. Adapter lens (f=400mm) in front of the Acer lens.
Annoying is during, respectively after the projector calibration that I have to wait more than 3 minutes until the small phase error window pops up. :(

I know that I still have to learn a lot. But as we say „Rome was also not built-up during one day“.

Walter Maximo
Attachments
Büste v Walter.jpg
User avatar
WalterMo
Moderator
 
Posts: 2363
Joined: Mon Apr 02, 2007 6:52 pm
Location: Braunschweig, Germany

Postby hal » Sat May 14, 2011 10:10 pm

Hello.

Concerning my old post on date Sat May 14, 2011 1:12 pm, as I wrote, I've done a mistake, now solved.
Here you can see a quite satisfying result about a dark object without coating. I've played a lot with PHT, Exposure and Gamma values.
The result seems promising.

MatM_D3_VeryDarkObject.jpg

The tests continued ... :wink:
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

Re: [Gallery] SL results

Postby MagWeb » Sat May 14, 2011 10:48 pm

hi Mattia

looks good.
You wrote, you made a mistake. May I ask which one?

Gunter
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm

Postby hal » Sat May 14, 2011 11:40 pm

hal wrote:But seems that if the white fringes are too bright, the noise is worse and artifacts appears. Simon and/or Sven: Do you think that can you fix this problem? when the light seems too much the mesh appear completely distorted... Ok, we must adjust our projectors, but not so easy with coloured objects with dark and light surfaces (for example, is impossible to scan stull like this (whitout coating), at the current state of art of D3). I've tried to scan a dark/light surface with the 4DDynamics's PicoScan and it works correctly, with the same projector and envirormental light (quite dark room). Ok, well, PicoScan create more noise on the black surfaces, but it can survey it and don't "burn" the white areas.
SOLVED - This was my fault! I've forgotten to modify the PHT value in relation to the new exposure value, sorry. Obviously at low PHT and high Exposure values, the software can't catch correctly all the patterns, so deformations appears :oops: .


I've increased the Exposure time but not the PHT :oops:
User avatar
hal
Moderator
 
Posts: 1527
Joined: Thu Mar 01, 2007 4:08 pm
Location: Cavaria con Premezzo (VA), Italy.

PreviousNext

Return to DAVID Development

Who is online

Users browsing this forum: No registered users and 1 guest

cron