CGAL users discussion list

[Jane Tournois <>]

Hi Alexandre,

it is strange indeed that the 50x50x50 works well and not the 300x300x300.

However, I have a (slightly unrelated to the issue) question : why are 
you trying to get such a precision?
I mean, is it necessary/desirable to generate a mesh that sticks to the 
voxels, and will be "staircase shaped"?

In my mind, generating a mesh with
facet_distance = sqrt(2) * voxel size
is enough to be accurate enough, and even better to get no staircase effect.

I'm still looking for an explanation why 0.1 on the 300 example is that bad.

Best,
Jane.

Le 05/10/2015 22:04, Alexandre Lapene a écrit :
> Hi Jane,
>
> Thanks for your answer.
> The size of a voxel is 1 for this data set. In reality it's much smaller and
> does work well if I use the real dimension, I think it's 10^-6.
> What is strange is the smaller sample I have provided 50x50x50 works well
> for facet_distance up to 10^-5  with same voxel size. Do you think there is
> a scaling issue? I will try with normalize sample to avoid precision
> effects.
>
> Best regards,
>
> Alexandre
>
>
>
> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/Isosurfaces-from-Gray-level-image-tp4661158p4661195.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.

--
Jane Tournois, PhD
R&D Engineer at GeometryFactory
http://www.geometryfactory.com/