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

Target Types

The place for all topics related to the 'Shapefusion' tool of DAVID.

Target Types

Postby MagWeb » Wed Dec 18, 2013 5:44 pm

Hi out there,

doing some Free alignment you try to align a partial mesh A to a target mesh B.
There are 4 types of meshes you can create with DAVID:
- A single partial mesh as you get it doing a single scan;
- A group of combined partial meshes (select all meshes to group in the list of files, do a RMB click on the list and click combine in the context menue). It is possible to uncombine such a group within a DAVID session to get back all the single scans;
- A reloaded fusion result of partial meshes;
- A reloaded bundle of partial scans as you get it saving a group of meshes to a single file. There's no simple way any more to get back all the partial sources.

To illustrate the different behaviours I always used the same mesh as the one to be aligned mesh.
The target meshes use the same mesh again to guarantee best fit. I only cutted this mesh into two halfs and moved one to get a little gap between them ( This is a similar constellation as you may get it between the first and the last aligned mesh of a 360° scan sequence due to alignment errors). This means that the to be aligned mesh should bridge the gap.

Targets1.jpg


(from left to right):
The first case is the most known one: For you align the source mesh to a single partial scan it fits perfectly to it but due to the displacement of the second half it cann't fit to this one. This second half simply is ignored.

The second column shows FreeAlign onto the partial meshes being grouped. As expected the source mesh does not fit to the group perfectly but tries to find an avaraged position to bridge the gap taking care about both grouped meshes.

Doing a fusion (without filling holes) remeshes the involved surfaces to be one. You might expect such a target (column 3) as being handled the same way as we saw before in case of grouped meshes. But it isn't so. DAVID alignes the souce perfectly to the bigger part and ignores the smaller as if there were two partial meshes as long as the halfs'-surfaces are not connected.

Finally a reloaded bundle of scans (4th column) is treated the same way as if it were a group. (Note: As said, you cann't uncombine!)

Maybe this is useful information

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

Re: Target Types

Postby Simon » Fri Dec 20, 2013 9:58 am

Hi Gunter,

Thank you for your interesting study! Your results match my expectations, except for column 3:
MagWeb wrote:Doing a fusion (without filling holes) remeshes the involved surfaces to be one. You might expect such a target (column 3) as being handled the same way as we saw before in case of grouped meshes. But it isn't so. DAVID alignes the souce perfectly to the bigger part and ignores the smaller as if there were two partial meshes as long as the halfs'-surfaces are not connected.

Indeed, in this case I "expect such a target (column 3) as being handled the same way as we saw before in case of grouped meshes". You said that this happens "as long as halfs'-surfaces are not connected". That's surprising, since the triangle connectivity is only used for calculating the surface normals and for computing the mesh boundary. So triangles play only a minor role during alignment. Therfore, I think it is questionable whether one can generalize the result in column 3 to other cases. However, an aligment with a fusion result should be performed only in rare exceptional cases.
User avatar
Simon
DAVID Developer
 
Posts: 423
Joined: Sun Jan 21, 2007 3:40 pm
Location: Braunschweig - Germany

Re: Target Types

Postby MagWeb » Fri Dec 20, 2013 12:25 pm

Thanks for your feedback Simon,

I was aware of the dominance of the vertices for the alignment. Therefor I was surprised by the column3 behaviour too and did this several times in different constellations: same result. I 'm gonna do some further experiments if this happened accidentally

@ Aligning to a fused target:
This is not ideal sure, but necessary if you are working on big projects which need more than a few partial scans. I had complicated tasks were I needed up to 100 (maybe some more) partial scans at a high resolution and system resources are limited to a certain point.
Another thing (and that's the point were I recognized the column3 thing): A reloaded, rough, lowest res fusion (of only as much partial scans needed to describe the object's overall shape) can be used as a skeleton to align the whole number of scans to a single target. This is quicker than aligning to a growing group of meshes. After aligning towards the skeleton the skeleton is deleted.
I'm gonna post this strategy more detailed when I found out more about the column3 behaviour.
MagWeb
Moderator
 
Posts: 2499
Joined: Wed Jul 18, 2007 8:48 pm


Return to Shape Fusion

Who is online

Users browsing this forum: No registered users and 1 guest