Isosurface evaluation without introducing any smoothing

Hi to everybody,

I've noticed with some simple tutorial example that the isosurface function introduces some kind of smoothing in my matrix before starting the calculation of the isosurface. For my purposes this is very problematic, I rather prefer to see a surface divided in several disconnected pieces than a compact and smooth one. Is there one different algorithm that evaluates isosurfaces without introducing any kind of smoothing? Do anyone knows how may i solve my problem?

Thanks in advance to anyone answering,

Fabio

Answer 1

[Sean de Wolski]

An isosurface has to interpolate by nature in order to calculate triangles.

What are you trying to do? You might be able to call patch directly if you are just looking for a visualization.

Answer 2

[Fabio]

I've created a 3D scalar field and I would like to identify connected areas of space that have a value higher than a certain threshold...

I could simply select all the cells that fulfill the requirement directly from my original matrix using a very simple command(matrix>threshold), but I liked the idea of extending a little bit more the selection by introducing some kind of 'approximation', that is, in my case, the added volume given by the presence of triangles connecting the cells with the same isovalue compared to the simple 'sum' of cells. My problem is that the smoothing introduced with the regular 'isosurface' function makes this approximation a little bit too strong and I’ve experienced the merging of two (or more) areas of my space that, with a proper isovalue, should be separated.

I hope I made myself clear... but I've to admit that I'm not quite sure.. ask for more specification if needed.

Thanks again for the support.

[Sean de Wolski]

Then perhaps the simplest method would be to come up with a better "isovalue" development algorithm. I.e. rather than fighting with isosurface work on making the values of your 3d matrix more distinct so that the isovalue threshold has more meaning...

[Fabio]

Probably you're right... btw i found this DelaunayTri function that maybe can help me...i'll try to use it. Thanks for your help!