Weird Laser Scan Reconstruction Problem

Follow

Tim B

• February 16, 2018 02:25

Got a weird problem with laser reconstruction where a small reconstruction region works fine, but  a bigger reconstruction region results in a mostly empty model.

Here is the process:

• Make a bunch of scans with the BLK360
• Register those in Recap Pro on PC
• Export e57 with all the scans from Recap Pro with Complete option (as opposed to 3D only)
• Import into RC (Registration = Exact, Geofenced = True, Color, Noise Free)
• Align 
• The result is like this:
• 
• Then, if I define a small reconstruction region (as seen above) and reconstruct in Normal Detail, I get the following. Which is what I would expect. 
• 
• However, if I set a large reconstruction region, as below:
• 
• Reconstruct in Normal Detail. Results in an empty model:
•  

Any thoughts on this? Any settings that may be causing this?

Thanks

0

• 

  Cabral

  • February 16, 2018 21:30

  Hi Tim,

   

  Try changing the "minimal distance between two vertex" (under mesh calculation) to a value different than 0. Check with 2mm for instance

  It looks like the same problem described here:

  https://support.capturingreality.com/hc/en-us/community/posts/115003082232-Laser-and-Photo-Reconstruction?page=1#community_comment_115001376172

   

  0

  Comment actions Permalink
• 

  Götz Echtenacher

  • February 18, 2018 09:29

  That's the one, thanks a lot Cabral!

  Tim, did that do the trick for you as well?

  0

  Comment actions Permalink
• 

  Tim B

  • February 20, 2018 05:57
  • Edited

  Hi Cabral and Gotz,

  Thank you for the reference. 

  I've actually read the post you linked (along with a bunch of others) while trying to figure this out.

  I saw the comment about "minimal distance between two points" but missed the distinction between that and "minimal distance between two vertices" !!  I was sure that everything was as you described in that post :)

  Did you just mess with all the settings until it worked?

  0

  Comment actions Permalink
• 

  Götz Echtenacher

  • February 20, 2018 07:10

  Hi Tim,

  the first one (points) has an extensive tooltip. Here, you set the minimum distance of the points in the scan, whereas with the latter (vertices) you define the density of the mesh.

  0

  Comment actions Permalink
• 

  Götz Echtenacher

  • February 20, 2018 07:12

  And if that doesn't help, then there is always the Bug Report section.  :-)

  Some people are really reluctant to use it, which is better than use it carelessly. But ultimately, that's what we are here for - RC is still officially in beta!  ;-)

  0

  Comment actions Permalink
• 

  Cabral

  • February 20, 2018 10:44

  Hi Tim,

  In my case after I changed the minimal distance between two vertex to some value (usually 2mm but also used other values) it worked everytime and in different projects.
  Today I actually got a reply from the RC team saying that they were able to replicate the bug and they will be looking into it.

  0

  Comment actions Permalink
• 

  Tim B

  • February 21, 2018 03:38

  Thanks you both for all the help and info!

   

  A new issue popped up. 

  I've been experimenting with various Min distance between points/vertices and the model gets generated ok now. Additionally, there is a visible difference between models with lower vs higher min. distance.

  However, for models that are generated with smaller min distance, there is a texturing issue. The texture step fails and a checkerboard pattern appears instead of expected texture.

  Here is a model generated with at 0.005. It textured fine:

  

  Here is the same thing generated at 0.002 distance, then attempted to texture:

  

  I'm guessing, as usual, there is something I'm missing. Searched for  checkerboard texture issues, but didn't find any. 

  Any thoughts? 

  0

  Comment actions Permalink
• 

  Götz Echtenacher

  • February 21, 2018 17:04

  Hi Tim,

  the checkerpattern you see is actually the blank unwrap or rather a dummy texture. If this behaviour is reproducable, it sounds like another bug to me.

  What differences do you see in models reconstructed with different min. distance? I assume you are not just talking about the difference in detail or density, right? Are there larger parts, that will be reconstructed differently?

  0

  Comment actions Permalink
• 

  Tim B

  • February 21, 2018 17:39

  Gotz,

  For different min. distance, I was talking about detail density. I meant to say that the min. distance setting does have an expected effect on the model/feature density and trig count. 

   "Are there larger parts, that will be reconstructed differently?"

  Not sure what you're asking. Do you mean whether larger parts will be different with different min distance setting? 

  0

  Comment actions Permalink
• 

  Götz Echtenacher

  • February 22, 2018 07:11

  Hi Tim,

  Ah I understand. Yes, I meant if there are for exapmle different bubbles in some tricky areas or somesuch, but you've made it clear already.

  0

  Comment actions Permalink

Please sign in to leave a comment.