Anisotropic API on required edges

Hi Algiane and Luca,

we are starting to look at the anisotropic API of mmg and we are having difficulties to impose the correct metric on required edges (2D only for the moment).
This is easily visualized on the attachment.
The initial mesh respect (left) perfectly the required anisotropic metric but when we remesh while setting the boundaries as required, an isotropic metric is used on the boundary points (right).
Is there a fundamental reason for that or can it be fixed easily?

Thanks for your help!

anisoAPI.pdf (1.2 MB)

Hi Ghislain,

For now it is the expected default behaviour of Mmg along required entities: at required points we impose a length that is computed as the mean of the length of the edges that are incident to the point. Then this size is slightly propagated due to the required gradation. You can forbid this behaviour (if you are sure that your metric doesn’t conflict with the required edges) using the -hgradreq -1 -nosizreq parameters.

The same issue occurs in 3D.

I hope that I will have the time to change that behaviour in the next month: the idea is to replace the current computation by the computation of an anisotropic mean edge length.

I hope that it will help,
Best Regards,


Thanks a lot for your answer Algiane!

Please let us know when the new anisotropic mean edge length is available.