Forum

Author Topic: Limits on polygons/face count for mesh?  (Read 7526 times)

saproling

  • Newbie
  • *
  • Posts: 4
    • View Profile
Limits on polygons/face count for mesh?
« on: May 19, 2014, 11:56:19 PM »
Hello,

I've been trying to increase the number of polygons in the mesh generated from my dense point cloud, but keep hitting a limit. The low setting will generate ~6 million polygons, the medium setting will generate ~12 million polygons, and the high setting SHOULD generate ~20 million. However, for some reason I can't generate more than ~17 million even when I specify the number of polygons to be much more. I have over 100 million points in the dense could to work with, though. Does anyone know why I can't seem to generate more than ~17 million polygons?

Thanks,
S

Marcel

  • Sr. Member
  • ****
  • Posts: 309
    • View Profile
Re: Limits on polygons/face count for mesh?
« Reply #1 on: May 20, 2014, 03:57:21 AM »
Which meshing method are you using,  heightfield or arbitrary? I can imagine that heightfield has a minimum grid size which would limit the number of polygons you get (I haven't tested this so I am not sure)

If you set the number of polygons to 'custom' and then fill out 0, Photoscan will build the mesh with the maximum number of polygons (in fact it builds the mesh with max polygons  anyway,  so there is no reason to always set it to 0 and reduce the polycount afterwards,  you will be more flexible. )

saproling

  • Newbie
  • *
  • Posts: 4
    • View Profile
Re: Limits on polygons/face count for mesh?
« Reply #2 on: May 21, 2014, 05:14:16 AM »
I've been working with height field. Setting the custom to 0 also resulted in ~17million polygons. I think I'll have to export my dense points and use another program to generate a TIN.

Marcel

  • Sr. Member
  • ****
  • Posts: 309
    • View Profile
Re: Limits on polygons/face count for mesh?
« Reply #3 on: May 21, 2014, 09:12:35 AM »
Try the arbitrary meshing method. I think the heightfield method is using some kind of grid with a fixed resolution (maybe based on the size of the region?)

Arbitrary meshing will be much slower though,  and takes more memory. But at least you will know what the problem is.