1

Topic: deformation test

I tried to assess scanning deformations that happen going from the center of the table towards the outside of the table. For that I choose a flat object, that is a very small book size 25mm x102mm x 210mm. I covered the book with millimeter paper that is paper covered with smaller 1mm squares and larger bolder 10mm squares. What can be easily perceived in the scan are the 10mm squares. Obviously all lines on the paper are straight. There is a rubber band that keeps the paper onto the book and this also is straight.

Results are shared here https://drive.google.com/open?id=0B8AL6 … lp5OTBpWWc

You'll find:
1 a .ply file with the scan at default high resolution
2. a photo of the object and the scanner
3. the laser test.
4. a snapshot of the scan that is faster to see than the whole scan

You'll see that the lines are not much deformed at a height of about 90 mm from the plate which is the height of the camera with respect to the table plane. But the lines are deformed above and under that height. The more you move from the camera level the more lines are deformed.

This deformation seems very serious since it is vary large. So I wonder if there is a fix. I wonder if my calibration is wrong, but the enclosed test seems ok. Or not?

Also on the lower right part of the snapshot  there is a part that for some reason was not scanned at all. But this has nothing to do with deformation I suppose.

Best
Fran

2

Re: deformation test

Anybody tried to repeat this test?  It is very easy, you put a flat object, like a small book,  standing on the plate. You cover the object with a paper having many straight horizontal lines covering the whole object and you scan it. That is all. It is a 10 min job.

But you'll have an idea of the deformation induced by your scanner. If your scanned straight lines are no more straight, but bended you can even measure how much they are deformed.

Best
Fran

3

Re: deformation test

That is much more deformation that you should see.  You calibration is surely off.  Can you post a picture with the calibration item centered on the table as described in the instructions?  Also, make sure that you are running the latest version of the software and perform the "laser calibration" as instructed in the instructions.

- Uriah

4

Re: deformation test

hairu526 wrote:

That is much more deformation that you should see.  You calibration is surely off.  Can you post a picture with the calibration item centered on the table as described in the instructions?  Also, make sure that you are running the latest version of the software and perform the "laser calibration" as instructed in the instructions.

- Uriah

Thanks Uriah for your response.

the photo you requested is added at the same folder: https://drive.google.com/drive/folders/ … lp5OTBpWWc with the name: calibration item centered. Anyway the vertical line on the calibration item is centered with the hole at the center of the table and the calibration item is parallel to the plane of the  arms of the scanner.

yes software was updated. The update went through without errors and then told that there were no more new updates.

Anyway, just for test purposes, it would be interesting if anybody could post the result of a scan of a vertical sheet, large 100mm, high 200mm and with many horizontal lines on it.  To me it seems an obvious practice, particularly for a self made product.
Best
Fran

5 (edited by C_D 2016-08-06 12:11:48)

Re: deformation test

I am keen to do this test on my scanner. I will try and make some time for it and post my results.

I have thought a lot about what other calibration procedures could be done with the software. I have done a little machine vision work before and we were correcting for camera lens distortion and all sorts. By scanning items of known size you can learn a lot about the accuracy of the machine.

6

Re: deformation test

Thanks Fran.  Can you post the picture of the calibration item as seen from the scanner's camera so that the red calibration lines are visible?  Also, make sure you click the "Calibrate Lasers" button while the calibration item is oriented on the table correctly.

7

Re: deformation test

hairu526 wrote:

Thanks Fran.  Can you post the picture of the calibration item as seen from the scanner's camera so that the red calibration lines are visible?  Also, make sure you click the "Calibrate Lasers" button while the calibration item is oriented on the table correctly.

the lasers photo is uploaded to the same folder above. The laser light is a little bit on the left on the upper part because when I tight the screw it moves so it is very difficult to align it correctly. However the deformation of the above test is so huge and symmetrical in the upper and lower part, that I feel it is quite independent from this lack of accuracy.

Anyway nobody posted any test yet!
Best
Fran

8

Re: deformation test

Hello,

to simplify the test, a sample of graph-paper is enclosed. Just print it, wrap it around a small book or small flat object (about 100 mm x 200 mm). A rubber band is enough to keep all together. Put it standing on the scanner plate and scan it.

Best
Fran

9

Re: deformation test

More than two months passed since I opened this thread.

Nobody posted a result of this very simple test. Nobody replies.

I imagine this may mean nobody is able to get better results. But if this is the case this scanner is just a prove of concept, but nothing you can work with for any type of production. An the declared precision is not at all true.

On the other hand fixing the software should be possible. But if nobody replies and nobody is interested....

10 (edited by tonyno 2016-12-11 14:15:58)

Re: deformation test

I finally have my scanner built and functioning. Working on this now.

Update: There is no file for the paper above, but I just found this page for generating something.

https://incompetech.com/graphpaper/leftborder/

11

Re: deformation test

Here are my files. Not sure what's going on with the left sides of objects. Maybe due to using the right laser only (default)?

https://drive.google.com/open?id=0BzWjC … 0RRcXV1dkE

12

Re: deformation test

tonyno wrote:

Here are my files. Not sure what's going on with the left sides of objects. Maybe due to using the right laser only (default)?

https://drive.google.com/open?id=0BzWjC … 0RRcXV1dkE

No idea, what happens on the left side, but your lines are not deformed as mine.

13

Re: deformation test

Well, that's good for me. tongue

Any official help here?

14 (edited by knowack 2017-03-04 03:42:41)

Re: deformation test

I'll play.

I'm not sure what is causing the problems you're experiencing, but here's my deformation test:
https://www.dropbox.com/s/d8z6vl96v4vpx09/stress%20test.png?dl=1

This is a piece of graph paper wrapped around a bottle of airbrush thinner, secured with two rubber bands.  It was scanned using the "Default" preset that came with the software, except that I'm using both lasers.  So, that means the parameters include 1.9Megapixels, 800 frames per revolution, Auto camera exposure time, medium noise removal, and medium image threshold mode for laser detection.

The squares of the graph paper are .25"; the top rubber band is approximately 120mm from the turntable.

There has been no post-scan cleanup whatsoever.  The lines appear straight and even.  I've used the Matter&Form scanner, and the Atlas3D.  With both, the secret is CALIBRATION and LIGHTING.  IMO, the Atlas3D does a better job.

15

Re: deformation test

Can you do a rectangular object per the OP? Round seems to be a cheat.

16 (edited by knowack 2017-03-05 00:09:25)

Re: deformation test

I wrapped the graph paper around a box of 9mm cartridges:
https://www.dropbox.com/s/jkv1f3vvkvk7gmg/stress%20test%202.png?dl=1

Same settings as above.

17

Re: deformation test

What are your 'Camera Y', 'Camera Z', 'Right Laser X' and 'Left Laser X' settings?

18

Re: deformation test

I started with Default and also tried two lasers.

From the log of my last scan...

setup.cameraLocation.y=82.55
setup.cameraLocation.z=266.7

setup.leftLaserLocation.x=-123.952
setup.rightLaserLocation.x=123.952

19 (edited by knowack 2017-04-25 22:12:25)

Re: deformation test

tonyno wrote:

I started with Default and also tried two lasers.

From the log of my last scan...

setup.cameraLocation.y=82.55
setup.cameraLocation.z=266.7

setup.leftLaserLocation.x=-123.952
setup.rightLaserLocation.x=123.952


Ok, there are some differences between our settings.  This is from my log:

setup.cameraLocation.x=0
setup.cameraLocation.y=82.55
setup.cameraLocation.z=260.35
setup.leftLaserLocation.x=-133.35
setup.leftLaserLocation.y=82.55
setup.leftLaserLocation.z=260.35
setup.rightLaserLocation.x=133.35
setup.rightLaserLocation.y=82.55
setup.rightLaserLocation.z=260.35

In the Atlas3d, under Settings | Setup, my units happen to be in inches:
https://www.dropbox.com/s/430upofj50ejdiu/Settings.PNG?dl=1

At one point, in a effort to tweak for optimal results (I'm scanning matchbox cars and wargame figures), I made measurements of the Atlas3D's actual geometry.  I found that the Camera Z physically measures to be about 10.6"...approx 269mm.  When I adjusted that setting, my point cloud went haywire, so I put it back.  I'm not sure where the notional Zero Datum line for Camera Z falls within the geometry, but it is not quite the physical location.  All of my other measurements matched the settings.

Your Camera Z is set to 10.5"; try dialing that back to 10.25" and see if your results improve.  Then consider adjusting your laser left and right Z...you're at 4.88", and I'm at 5.25".

I have the laser-cut acrylic frame, so its dimensions are consistent.  If you're using a 3D printed frame, you may need to tweak your numbers.  I haven't gone back and attempted to further refine my dimensions, so you may find settings that work better than mine!

20

Re: deformation test

Thanks. I also have the acrylic frame. I will test this.

Screenshot...

https://dl.dropboxusercontent.com/u/6626323/atlassettings.jpg

21

Re: deformation test

How can i fix this?
The original model is completely square.

http://soliforum.com/i/?yEQYOIG.png

22

Re: deformation test

Make sure that the object is not moving around on the table.  Are you using a non-slip surface on the table.

23 (edited by Guffer 2017-04-24 18:50:42)

Re: deformation test

Using the rubber mat following the kit.
The building is 13x12 cm size.
Looks like its going to implode tongue

http://soliforum.com/i/?TvdzWkj.png
http://soliforum.com/i/?wLtrJEV.png

Post's attachments

2017-04-24 19_55_42-MeshLab_64bit v1.3.3 - [Project_1].png
2017-04-24 19_55_42-MeshLab_64bit v1.3.3 - [Project_1].png 322.04 kb, file has never been downloaded. 

2017-04-24 19_56_17-MeshLab_64bit v1.3.3 - [Project_1].png
2017-04-24 19_56_17-MeshLab_64bit v1.3.3 - [Project_1].png 300.24 kb, file has never been downloaded. 

You don't have the permssions to download the attachments of this post.

24

Re: deformation test

That's so weird.  It looks like a reversed "fisheye" distortion.  When you view the turntable in real time, do you see this?

I'm pretty sure this is related to the camera itself.  Is this the original RaspiCam, or something else?  Did you add some after market lens to this? 

I know there are some settings that can be done on the camera from within Raspbian, but I don't know hat they are. Maybe Uriah can help.

To print or, 3D print, that is the question...
SD3 printer w/too many mods,  Printrbot Simple Maker Ed.,  FormLabs Form 1+
AnyCubic Photon, Shining 3D EinScan-S & Atlas 3D scanners...
...and too much time on my hands.

25 (edited by Guffer 2017-04-25 05:30:36)

Re: deformation test

Its not something with the camera, it looks right when viewed via camera.
Its the original PI camera, just recieved the full kit 5 days ago, and havent been able to get a nice scan yet.

http://soliforum.com/i/?D74mAy2.png