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

DAVID 5 Beta

Discussions about latest DAVID developments, beta versions...

Re: DAVID 5 Beta

Postby micr0 » Fri Sep 02, 2016 5:06 pm

Where is the link to the release of David 5.0.4? It isn't available under downloads on the David support web page.
µ
micr0
 
Posts: 315
Joined: Wed Mar 19, 2014 3:54 pm
Location: NYC

Re: DAVID 5 Beta

Postby Sven » Tue Sep 06, 2016 3:31 pm

Hi,

We had an unexpected problem in our build process which broke 5.0.4 in several ways , so we had to pull it back. :(

We have now fixed that, and released it as 5.0.5. This version we have tested more thoroughly than ever. Please give it a try! Especially in the "Quality" profile, you should get very low distortions for Dual Camera setups.

Sven
User avatar
Sven
DAVID Developer
 
Posts: 1541
Joined: Mon Jan 22, 2007 8:59 am
Location: Braunschweig, Germany

Re: DAVID 5 Beta

Postby micr0 » Tue Sep 06, 2016 9:40 pm

I am not sure if I'm doing something wrong, but I am having a problem with David5 but not having a problem with David 4. Here's what I have:

This is the camera/projector/calibration panel set up:

5cal6.jpg


These are the 300mm calibration panels I mad out of 6mm glass panels and a CNC cut aluminum stand I made to hold them. As you can see they are at 90deg. (I have also verified this with a Romer CMM that is accurite out to 3 decimal places.

IMAG1387.jpg



Here is what the camera sees/the results of the calibration process in David 5:

5cal1.jpg
µ
micr0
 
Posts: 315
Joined: Wed Mar 19, 2014 3:54 pm
Location: NYC

Re: DAVID 5 Beta

Postby micr0 » Tue Sep 06, 2016 9:46 pm

After calibration and with out changing a thing i scan the calibration panels with David 5, which looks like this:

5cal7.jpg


This looks good until you rotate the scan so that you are looking normal to the edges of the scan. Then you can see the distortion in the form of the planes of the two calibration panels not being 90deg to each other.

5cal2.jpg


5cal3.jpg
µ
micr0
 
Posts: 315
Joined: Wed Mar 19, 2014 3:54 pm
Location: NYC

Re: DAVID 5 Beta

Postby micr0 » Tue Sep 06, 2016 9:55 pm

Lastly I quit David 5 and boot up David 4.5.3

Again Calibrate, then scan the calibrate the calibration panels. This time they are perfectly 90 degrees to each other.

5cal4.jpg


The only difference is that one scan was done with David 5 and one was done with David 4. I didn't touch or move anything at all (aside from the mouse on the desk to click).

Notice that the results returned by the calibration in David 4 are different:

Davbid4 cal.jpg
µ
micr0
 
Posts: 315
Joined: Wed Mar 19, 2014 3:54 pm
Location: NYC

Re: DAVID 5 Beta

Postby micr0 » Tue Sep 06, 2016 11:26 pm

*UPDATE*

I flipped the calibration corner to a concave configuration and now I seem to be getting nearly perfect calibrations in David 5. The scans of the panels are almost perfectly square too, and the calibration specs (camera angle etc.) are now almost Identical to the figures returned by David 4 (still slightly different though). It would seem that David 5 doesn't like outside calibration corners.
µ
micr0
 
Posts: 315
Joined: Wed Mar 19, 2014 3:54 pm
Location: NYC

Re: DAVID 5 Beta

Postby Sven » Fri Sep 09, 2016 3:52 pm

Thanks Micr0,
DAVID5 should "like" the "negative corner" as much as DAVID4, but I will investigate!
Sven
User avatar
Sven
DAVID Developer
 
Posts: 1541
Joined: Mon Jan 22, 2007 8:59 am
Location: Braunschweig, Germany

Re: DAVID 5 Beta

Postby micr0 » Tue Sep 13, 2016 12:34 am

Sven I spoke too soon. :( :(
I tried working with David 5 today and even with the concave panel was right back where I started with the scan of the cal panels that were off. I double checked with david4 and again not touching anything, calibrated and then scanned the panels. I exported one scan of the panels from David 4 and one scan of the panels from David 5 in .obj format and brought them in to Geomagic. The David 4 scan read 90.061 degrees.

David 4 cal panel.jpg


Where as the David 5 Scan measured 88.0332 deg

David 5 cal panel.jpg


Again the camera angles relative to the projector that are reported after the calibration sequence are different.

Here are a few things I noticed:

In David 5 after calibration, when the black and white checkerboard is displayed connecting the dots on the calibration panel. the corners of many of the squares are not in the center of the black dots.

I can't really tell but it looks like the field of view is slightly different between the two versions. Does David 5 see the video stream differently or do I have something set wrong?

output_prr3x8.gif


Its a subtle difference but there is a difference, and it may be worth 2deg of error?

What would cause 2 different versions of the software to scan, process and render 2 different calibrations?

I really hope we can get to the bottom of this as the scans from David 5 are cleaner than David 4, and I want to invest in a second similar camera (David 5 really doesn't seem to like two different format cameras) but I can't justify that expense while I can't get good scans from David 5.

Hal (if you are listening)
have you had any luck? IIRC you were having similar problems to mine.
µ
micr0
 
Posts: 315
Joined: Wed Mar 19, 2014 3:54 pm
Location: NYC

Re: DAVID 5 Beta

Postby micr0 » Tue Sep 13, 2016 1:37 am

Sven wrote:Hi all,

Thank you for your reports. We have done a lot of tests, and indeed, in some setups the calibration can be somewhat off, so that scans of left and right camera don't fit together (Extended View mode only).

We have improved calibration for two cameras. It is now much more robust especially when the calibration corner is not built precisely, or when the camera view direction is not ideal.
For those who have seen that problem, please try our new beta version: http://downloads.david-3d.com/software/ ... _beta1.zip
The relevant new setting is at Advanced Settings -> Calibration -> OptimizeMultiCameraCalibration (on by default)

Thanks
Sven



THIS WAS IT! This setting is on by default but should NOT be on if you are using only one camera!. This is what was throwing off the calibration.

ARRRRGE the amount of time I spent on this. Ach du lieber! (My grand father used to say this a lot when i was around him and now I think I know what it means :D)

I'm going to try a few scans and see how it works. Then I'm liberally applying beer therapy for the rest of the evening.

For those that don't follow: If you are using only one camera with David 5 make sure "OptimizeMultiCameraCalibration" is off.
µ
micr0
 
Posts: 315
Joined: Wed Mar 19, 2014 3:54 pm
Location: NYC

Re: DAVID 5 Beta

Postby micr0 » Tue Sep 13, 2016 4:14 am

:D

2016-09-12_23-07-43.jpg



5 is working!
µ
micr0
 
Posts: 315
Joined: Wed Mar 19, 2014 3:54 pm
Location: NYC

Re: DAVID 5 Beta

Postby Sven » Tue Sep 13, 2016 10:14 am

Hi Micr0,
Good to hear!
"OptimizeMultiCameraCalibration" indeed also does some optimization when there is just one camera. That may be dangerous... We'll discuss and probably switch it off for 1 camera.
Do you now also get good results with convex "negative" corner calibration?
Thanks!
Sven
User avatar
Sven
DAVID Developer
 
Posts: 1541
Joined: Mon Jan 22, 2007 8:59 am
Location: Braunschweig, Germany

Re: DAVID 5 Beta

Postby micr0 » Wed Sep 14, 2016 2:34 pm

Sven wrote:Hi Micr0,
Good to hear!
"OptimizeMultiCameraCalibration" indeed also does some optimization when there is just one camera. That may be dangerous... We'll discuss and probably switch it off for 1 camera.
Do you now also get good results with convex "negative" corner calibration?
Thanks!
Sven


I haven't tried the negative corner again yet. I can next time I'm scanning. FWIW that transmission has been my test subject for the last few months. It is as deep as it is wide, has some fine casting marks (detail) and difficult negative spaces etc. Yesterday I got time to do a complete scan of the trans (54 scans) and the results are fantastic! There are a number of machined faces that I can easily measure and compare to the measurements returned by points in scan data. My favorite points are two parallel machined faces that are 295.71mm apart (measured by CMM). In David 5 these points measures 295.57mm a deviation of ~139.7µ. Amazing!

And you say 2 cameras is more accurate?
µ
micr0
 
Posts: 315
Joined: Wed Mar 19, 2014 3:54 pm
Location: NYC

Re: DAVID 5 Beta

Postby drndadoo » Wed Sep 14, 2016 2:40 pm

fantastic results micro, do you want to share your scans here in forum please? I would love to see more screenshots :)
drndadoo
DAVID Support
 
Posts: 194
Joined: Tue Mar 24, 2015 3:09 pm

Re: DAVID 5 Beta

Postby Sven » Wed Sep 14, 2016 3:19 pm

Happy to see that, after you had so much trouble... :D

micr0 wrote:And you say 2 cameras is more accurate?

There is no easy answer. Let me try: In ideal conditions, Dual-Camera setups are not more accurate than one-camera setups. However, in real life, the following effects can and do happen:
    - Thermal drift in projector --> Dual Camera is immune
    - limited quality of most projector lenses (distortions) --> Dual Camera is immune
    - "difficult" / deviant reflection of projected pattern on some object surfaces, e.g. when textured --> Stereo is more robust against that
    - etc.

Sven
User avatar
Sven
DAVID Developer
 
Posts: 1541
Joined: Mon Jan 22, 2007 8:59 am
Location: Braunschweig, Germany

Re: DAVID 5 Beta

Postby micr0 » Wed Sep 14, 2016 3:30 pm

Sven wrote:Happy to see that, after you had so much trouble... :D

micr0 wrote:And you say 2 cameras is more accurate?

There is no easy answer. Let me try: In ideal conditions, Dual-Camera setups are not more accurate than one-camera setups. However, in real life, the following effects can and do happen:
    - Thermal drift in projector --> Dual Camera is immune
    - limited quality of most projector lenses (distortions) --> Dual Camera is immune
    - "difficult" / deviant reflection of projected pattern on some object surfaces, e.g. when textured --> Stereo is more robust against that
    - etc.

Sven


What about error correction/anomaly rejection?
µ
micr0
 
Posts: 315
Joined: Wed Mar 19, 2014 3:54 pm
Location: NYC

PreviousNext

Return to DAVID Development

Who is online

Users browsing this forum: No registered users and 1 guest