New DAVID Version 2.3.4

[Sven]

Hello everybody,

We have just released DAVID Version 2.3.4!
As you see, it's not 3.0, and not 2.4, just 2.3.4, meaning it's not a HUGE step, but anyway, there are some reasons for everyone to update!
http://www.david-laserscanner.com/?section=Downloads
If you own the USB Edition, please see here:
http://www.david-laserscanner.com/wiki/ ... update_usb

So what's new?
The most important thing is that we have increased scanning speed with less CPU usage. Depending on your hardware, versions 2.2.x were significantly slower than version 2.1c. Now we are nearly as fast again. Of course we are still not faster than your camera (e.g. 5 fps in 1600*1200 with most webcams). Thanks especially to Mattia for reporting the problem!
Main reason was that in versions 2.2.x we had switched to new c++ compiler version 9. Now we are back at v.8. Also there were some parts of our code to be optimized...

Calibration is also improved:
- If it fails, it shows you which markers have been found, so you can see where you have to optimize/clean the camera image.
- If it succeeds, it shows the calibration error (difference between model and reality) as short green lines. For each marker, there is a small difference between its real pixel coordinates (found in the camera image) and its model position (3d coordinates projected into the camera image according to camera model parameters). These lines are stretched by factor 10! So if you have a 10 pixel long line, this means the calibration marker is 1 pixel away from where the camera model thinks it should be.

Scanning Without Background:
Works better with imperfectly constant laser speed. Unfortunately even with the best hardware, it seems the laser motion speed is not perfectly constant. This results in a distorted scan ("macro waves"). Thanks to Greg, Gunter, Mattia, and Walter for their tests and reports! Now this problem should not occur any more. Theoretically, you can have ANY (irregular) laser motion as long as it can be perfectly reproduced! See new Advanced Settings "UseRememberedPlanePoses" and "UseModelOutsideReferenceRange". I will explain in more detail later what this is about.

New function in Shapefusion: "Save all scans in separate files". So you can save your current Shapefusion workspace: all scans in their current position, with texture, as separate files. Thanks to everyone who came up with this idea!

And some more, see here:
http://www.david-laserscanner.com/wiki/ ... se-history

If you experience any problems with the new version, please let us know!


It's never too early to wish you a

Nice weekend!
Sven

PS: Now we are very optimistic that the next version will be 2.4 - including the new camera grabber interface that is compatible to more different cameras and probably less CPU consuming... Please be patient.

[hal]

Hello Sven,
Great!
I'm impatient to try the latest version, without panels.
Is rare that a team's software works so quickly and excellently, for their product: many "big" softwares are updated slowly during many years and not with so much new features, as is for David.
Compliments!

Best regards and good work for the 2.4 version,
Mattia

p.s.: the "Save all scans in separate files" feature, is my favourite: my workflow speed up very fast, now... if in the past this saving operation take me many minutes, now the "saving" step is a pleasure, one click and go.

[MagWeb]

Hi Sven,

that´s not a small update, but a real important improvement for the scanning without background!
One can see the difference between estimation of the planes and using stored planes, when DAVID switches back to estimation for there are no stored planes any more...
great work.

Just some notes:
- For the logitech now there are much more resolution options, when you set the used resolution. Choosing a wrong one during calibration caused crashes at my system.
- A line that does NOT move is detected as a linear motion.
-- I found closing DAVID, that in this case the laser_motion.xml is changed to motionType linear. But if there are any planes stored in the xml, they are not erased. If one performes a scan of an empty corner with a rotating laser refering to the stored planes (rotating), but with the wrong motion_Type linear , the scanresult is a perfect plane! - funny
- David-laserscanner.exe misses its icon

BTW: I found that enabling the RightLight function in the logitech driver disables the settings of exposure and sensitifity

thanks
Gunter

[Sven]

Thanks for your quick feedback. I'll look into it...

[Godehardt]

Hello,
I just nows downloaded the new version and will use it in the afternoon and at the weekend!
Thanks for the continuous work on the software!

[cgdigg]

Nice update. Have posted this news on CGDigg:
http://www.cgdigg.com/story.php?title=david-version-2-3-4-released

[velleca55]

Thanks ... ...

[Sven]

Hello Gunter,

Thanks again for your remarks!

- For the logitech now there are much more resolution options, when you set the used resolution. Choosing a wrong one during calibration caused crashes at my system.

I have just updated the Logitech driver from 11.5 to 11.9. THIS gives new resolution optios, not the DAVID update. We didn't change anything about that. I have tried some of the "strange" resolutions, my system does not crash.
Note: When you choose a resolution that is not 4:3 but wider (higher), top and bottom (left and right) image regions are cut off.

- A line that does NOT move is detected as a linear motion.

That's true. But I think it's ok. It's a linear motion with speed 0. (It's also a rotation with speed 0). Will not appear in usual DAVID scanning. But if, for some reason, someone does this, the estimation is "correct" and can be used during Repetition even if the Repetition takes longer than the Reference Scan.

-- I found closing DAVID, that in this case the laser_motion.xml is changed to motionType linear. But if there are any planes stored in the xml, they are not erased. If one performes a scan of an empty corner with a rotating laser refering to the stored planes (rotating), but with the wrong motion_Type linear , the scanresult is a perfect plane! - funny

Are you sure? I cannot reproduce that. The list of planes is always updated. When I use a fixed laser during Reference, I get a list of the very same laser plane again and again. Maybe I misunderstood you?

- David-laserscanner.exe misses its icon

Yes, thanks, I have fixed that.

BTW: I found that enabling the RightLight function in the logitech driver disables the settings of exposure and sensitifity

Where is that? Sorry I can't see that setting...

Bye
Sven

[MagWeb]

Hi Sven,

thanks for looking into it!
I didn´t notice till now, that the 1.9 Driver ships this odd resolutions. In my case DAVID crashed using the mask/eraser function in calibration - I´m going to reproduce the problem and report.
[url]But if, for some reason, someone does this, the estimation is "correct" and can be used during Repetition even if the Repetition takes longer than the Reference Scan.[/url]
Thanks for the hint . Meanwhile I came to the same conclusion and did some tests already.

I didn´t pay further attention on the reported misbehavior. I´m looking into it this evening and report.

The RightLight function can be set for the Pro 9000 in the Logitech software.
Just a moment. I switch the partion and post a pic.

so long Gunter

EDIT:
so, here you´ll find the RightLight:

[MagWeb]

Hi again Sven,

seems the new res options of the Pro9000 simply provoke to play with them.

My DAVID crashes happen:

If I :
-"mask all" of any picture at lower res (maybe 320x240);
- AND try to switch to a higher res (maybe 640x480);

+no matter whether 4:3 or not,
+no problem occurs from higher to lower res

I checked the same action in 2.3.3. Here there seem to be no problems doing this (OK-to be admitted) silly workflow.

Gunter

[marting]

Hi to all,
I just wanted to ask a few questions:
A) I've just purchassed DavidPro 2.3.3 on USB and before I upgrade, I would like to know if the settings in the advanced settings are resetted to default when upgrading to to ver 2.3.4. I do not want to loose the settings that I've made here so as to have all indicators green (at the bottom of the window while scanning).
B) I have a liner laser movement ( up and down) with a fixed angle but have observed that the laser line angle indicator (at the bottom of the window) varries while scanning and at a point lights up from green to yellow. Does this mean that for some reason the scanning result is affected even though the laser line is actually fixed at approx 40 degrees at all times?
C) I am experiencing a considerable amount of scan variation between one scan and another, that is when I scan the object on one side then turn it about an axis to scan another area with some overlap so as to fit in like a jigsaw puzzle and be easier to locate one scan to another in the next step. Sometimes this variation is so much that David is not able to locate one scan to another at all. Is there something I am doing wrong or maybe some harware I need to change? The below list is what I am using / doing:
1 - Laser line thickness about 1.2mm.
2 - Slow linear movement motor driven up and down (approx 60mm per minute)
3 -Webcam 8Mpixel picure res (video 2Mb res at approx 9fps)
4 -Webcam setting black and white and all automatic setting disabled
5 - Scanning environment pitch dark (while scanning ony the laser line is visable)
6 - Tried scanning both matt and shiny objects (the latter coating with baby powder to take away the shiness)
7 - Also experimented with color shade of object being scanned - the best results I got was when scanning yellow objects, but still with scan variations.

I realy do hope to find a solution as I would be very discouraged not being able to achive an approximate result simpilar to many scans shown on this same forum!!!

Thanks
Martin G

[MagWeb]

Hi,

A: a new installation means a new advanced settings.xml file and also a new config.xml (here your settings and the calibration are stored). To save your old settings to the new version, simply copy your old advanced settings.xml and config.xml to the new folder. (Same as you do with your license files)
B: linear motion of the laser means: the laserplane is getting more and more parallel to the viewing lines of the cam the nearer it get to the cam. So I prefer a rotating motion of the laser, about at about the same distance cam/object as object/laser. This way you do the best you can to keep the intersectionangle almost constant.
C: You need at least 6 scans to get big enough overlaps to make an automatic registration of the partial meshes possible (the more, the better). If your meshes do not own the same topology DAVID will fail. So it has to made sure that your DAVID calibration and/or your reference planes are precise. A good test is to disable the background filter and scan your empty corner. The result should be perfectly flat and fit to the grey grid, so it can be seen from both sides all over. If it does not, check your setup.

Sure, the color an object owns affects the amount of laser light that is reflected. So green surfaces may cause trouble using a red laser and the other way round. Making the color a little paler (by using some powder) will give better results.

Gunter

[marting]

Hi MagWeb,
Thanks for your reply

A: a new installation means a new advanced settings.xml file and also a new config.xml (here your settings and the calibration are stored). To save your old settings to the new version, simply copy your old advanced settings.xml and config.xml to the new folder. (Same as you do with your license files)

I will save these files before updating.

B: linear motion of the laser means: the laserplane is getting more and more parallel to the viewing lines of the cam the nearer it get to the cam. So I prefer a rotating motion of the laser, about at about the same distance cam/object as object/laser. This way you do the best you can to keep the intersectionangle almost constant.

I prefer the liner over the rotating for 2 reasons: (1) the laser line angle is constant all the time and (2) at alater stage I will be scanning without the background so I am forced to have a liner movement instead. However I still do not understand why the laser line angle indicator at the bottom of the window is varying while scanning.

C: You need at least 6 scans to get big enough overlaps to make an automatic registration of the partial meshes possible (the more, the better). If your meshes do not own the same topology DAVID will fail. So it has to made sure that your DAVID calibration and/or your reference planes are precise. A good test is to disable the background filter and scan your empty corner. The result should be perfectly flat and fit to the grey grid, so it can be seen from both sides all over. If it does not, check your setup.

I have made repeated scans as you are saying but still having variations and the calibration corneris perfectly right angle, however I have never performed this test and will do so just to verify that it is ok.

On the otherhand, the reason I am a bit confused because from what I see in most of the settings and setup, it seems to be ok (as far as I can understand and from what I mostly read on the forum).

Anyway I will wait to recieve these focusable laser line generator and take it from there, in the mean time I would welcome andone's feedback.
Thanks
Martin G

[Ordibble Plop]

I prefer the liner over the rotating for 2 reasons: (1) the laser line angle is constant all the time

This is not actually the case because the angle is not calculated as the angle between the laser plane and the background panels, but rather the angle between the laser plane and the camera, i.e. draw a line from the laser to the panel and then to the camera and that is the angle being calculated.

Magweb provided a picture showing changing intersection angles for different setups (8 posts down) - viewtopic.php?t=588&highlight=intersectionangle

You can see from this that as a laser following a linear movement gets closer to the camera the angle becomes less and less.

[MagWeb]

Would be to add, that also for scanning without background, you are not forced to move the laser linear.
Also here: the rotating laser is the better choice, for the range the beams intersect the viewing angle isn´t limited by the mechanics.

See:http://www.david-laserscanner.com/forum/viewtopic.php?f=2&t=834
here I compare the rotating laser with a video projector. A linear moved laser would even be worse.

The only technic you are forced to use a linear movement is a scan using mirror reflections (see WiKI)