Hi 

Im trying to visualize n different sets of [x,y,z] data constituting n models of a surface in 3D space found by Inverse Monte Carlo. Im thinking about making a density plot of these n sets so as to indicate areas of larger probability for this surface. Does anyone know how to do this? I cant seem to find any 3D density method that applies. 

Also, im more than openminded towards suggestions of other ways of visualizing these data. Im aware that this particular visualization is difficult to make in 3 dimensions.  

Thanks in advance!

Are your models continuous surfaces, or just a collection of data
points?  What about using scatter3()?
Other options, possibly, (for indicating 3D densities in a 2D
picture):
http://www.ncgia.ucsb.edu/projects/Cartogram_Central/gallery.html

ImageAnalyst <imageanalyst@mailinator.com> wrote in message <03d52962-9e38-49e4-a692-f67af8724ebd@j2g2000vbo.googlegroups.com>...
> Are your models continuous surfaces, or just a collection of data
> points?  What about using scatter3()?
> Other options, possibly, (for indicating 3D densities in a 2D
> picture):
> http://www.ncgia.ucsb.edu/projects/Cartogram_Central/gallery.html

Each 'model' is a large number of points in 3D space, defined by [x,y,z] components so essentially their combined nature as surfaces can be disregarded in this discussion. I mentioned it in the case that anyone had any ideas for alternative ways of visualization. 

As I see it the Scatter3() command colorcodes the points in whatever way I prescribe. This is only half the solution, as I need that color to be determined by the density of points in that area. For a similar 2D case (points defined by [x,y] only) I used smoothhist2D (link below). Im considering if I could rewrite it to simply count number of points in 3D boxes rather than 2D boxes. 

Also I looked at the use of isonormals() as a possible solution: http://www.ks.uiuc.edu/Research/vmd/spotlight/images/voldata.jpg
does anyone have any experience with this?

Smoothhist2D: http://www.mathworks.it/matlabcentral/fileexchange/13352-smoothhist2d

Jakob:
I'm not familiar with those applications but it looks like they're
basically getting you a new image where the image values are related
to the density of points around a local region.  You can do this with
the convolution (conv() function), if you convert your scattered data
to a 2D array first.  Basically you sum together the values in a
sliding window.  More points in the window means a greater sum.  Then
you can just display that image (the one outputted by conv()).  For
the missing values where you have no x,y,z coordinate, I think you can
just have those locations be Nan's and those locations won't be
included in the sums.  This appears (to me) to be essentially what
they're doing.
-ImageAnalyst