Low resolution of meshes?

Question

Low resolution of meshes?

matraspe opened this issue 10 years ago · comments

Hi,

After successfully compiling and running cork (Windows 7 64) I was quite surprised to see the very coarse resolution of the objects: in the screenshot the result for the sample geometry (difference of ballA.off and ballB.off) is shown. Is there a setting of grid size or samples I have missed? Similarly coarse results occur for other custom meshes, too.

Thanks and regards,
Matthias

gilbo · Answer 1 · Sun Jul 27 2014 00:10:51 GMT+0800 (China Standard Time)

That's very strange. Can you tell me what order the coordinates your model uses are? Is it very far from the origin?

There is a slight quantization for the purposes of arithmetic, but it a should be at the 10^-6 scale relative to the largest coordinate value, so it should be unnoticable.

-- Gilbert

On Jul 25, 2014, at 1:12 PM, matraspe notifications@github.com wrote:

Hi,

After successfully compiling and running cork (Windows 7 64) I was quite surprised to see the very coarse resolution of the objects: in the screenshot the result for the sample geometry (difference of ballA.off and ballB.off) is shown. Is there a setting of grid size or samples I have missed? Similarly coarse results occur for other custom meshes, too.

Thanks and regards,
Matthias

�
Reply to this email directly or view it on GitHub.

matraspe · Answer 2 · Sun Jul 27 2014 03:57:28 GMT+0800 (China Standard Time)

Thanks for answering: actually, the depicted example uses the sample data included with cork (ballA|B.off). My own data is somewhat translated from the origin, but in the range of 50-150 units.

gilbo · Answer 3 · Sun Jul 27 2014 05:08:43 GMT+0800 (China Standard Time)

Ahh, I just noticed you're on Windows. In all likelihood there's something weird about how MPIR is being used that's breaking the code. This is very likely to be an arithmetic bug, so that's the obvious problem area.

-- Gilbert

On Jul 26, 2014, at 12:57 PM, matraspe notifications@github.com wrote:

Thanks for answering: actually, the depicted example uses the sample data included with cork (ballA|B.off). My own data is somewhat translated from the origin, but in the range of 50-150 units.

�
Reply to this email directly or view it on GitHub.

matraspe · Answer 4 · Sun Jul 27 2014 15:56:05 GMT+0800 (China Standard Time)

Yes, only MeshLab was running on Mac. Any chance to narrow it down? MPIR seems no fun... Do you have a prebuilt test binary for windows available? Thx

gilbo · Answer 5 · Mon Jul 28 2014 02:42:45 GMT+0800 (China Standard Time)

Did you verify that MPIR is working? I don't have a test binary ready. I can try to poke around my windows box later.

On Jul 27, 2014, at 12:56 AM, matraspe notifications@github.com wrote:

Yes, only MeshLab was running on Mac. Any chance to narrow it down? MPIR seems no fun... Do you have a prebuilt test binary for windows available? Thx

�
Reply to this email directly or view it on GitHub.