Functionality of Overlap setting in Refine Align

[BenjyvC]

I'm unclear how to use Overlap setting within the Refine Align tool. I see I'm not alone, this video at about 30:20 in poses a similar question, not clear how to think about what effect a low, mid, or high percentage overlap has when refining an already aligned pair of meshes or point clouds. I've seen the data mesh jump from being very closely aligned using point pair tool to scale jumping way out. How does the type data influence the outcome? I'm working variously with photogrammetry, scandata from terrestial lidar and from handheld scanner (Metrascan 3D from Creaform). I have to believe the quality of the data and the way it comes in must make a big difference in how CC sees and works with various scandata. Thanks for insights.

[daniel]

While the noise of the data shouldn't have too much impact on the Point-pair based alignment tool (see the other recent post on this subject), the ICP tool would have some hard time working in such conditions. And the overlap parameter will have a big impact, as it will tell CC to use only the close points, or more points, or even all the points from the 'to-be-aligned' cloud.

[BenjyvC]

Ah, so if the to be aligned cloud represents but half of the size of the target cloud, I'd use 50% overlap to constrain the search closer to the control points set in the to be aligned mesh, yes? Thanks.

[daniel]

Yes, either because only 50% of the aligned cloud is overlapping, or also more generally because 50% of the points don't match with the reference cloud for other reason (too high deformation, etc.)

[BenjyvC]

Clear, thanks. I'm afraid I'm having to visually align in MeshMixer, the noise in my photogrammetric model is evidently too high for CloudCompare tp see what to my eye looks clearly in common. It came sort of close, scale seemed about right, translation also, but rotation was off a good 20% on one axis, not sure if also on another.