Jonathan:

Thank you for your reply.

Your solution helped, and when I try it, it works. There are two concerns I have:

1) Is it possible to get the surface faces of the contours of different materials in the object?

2) Do you have more documentation on the matlab outout of cleaver? The nodes are the x,y,z positions and the cells are the elements with the node ids, right? For some reason, plotting the nodes shows points outside of the object. If I read the .ply file, the mesh looks alright. Do you know what could be the problem>

3) Do you, by any chance, have a Matlab tool to read and write NRRDs? I've been trying to write one, but it is not working yet.

Thanks for your help

On Mon, Feb 4, 2013 at 3:33 PM, Jonathan Bronson <bronson@sci.utah.edu> wrote:

Hi again Fernando,

In case it wasn't clear from my previous email, I've
supplied the 2nd material you need in order to create
a mesh with Cleaver. This corresponds to the option
(b) i described before. I've also attached the output
isosurface created by this input.

Cheers,
          Jonathan

-------- Original Message --------

Subject:	Re: Cleaver usage and problems
Date:	Sat, 02 Feb 2013 19:37:03 -0700
From:	Jonathan Bronson <bronson@sci.utah.edu>
To:	Fernando Quivira <f.quivira@gmail.com>
CC:	cleaver-request@sci.utah.edu, "D.Brooks" <brooks@ece.neu.edu>

Hi Fernando,

Thank you for the detailed system specs and problem description.
That's an excellent system and I don't forsee any issues running
Cleaver for future datasets.

The dataset you sent us doesn't work because it is a single material
file with a non-signed indicator function:

Cleaver supports non-signed indicator functions, but it also requires
at least two material files to produce a non-empty mesh. If a single
material file is given, an "inverse" indicator function is created
automatically by scaling the data by -1. However, this only works
for signed-indicator functions.

There are at least two ways you could alter your input to get your desired mesh.
a) You could modify test.nrrd such that values range from -0.5 to +0.5,
     instead of 0 to 1 as they currently do. Cleaver would then generate a
     suitable inverse mesh on the fly.
b) You could generate an inverse.nrrd file unique to your input. In this case,
     it  looks like swapping the 0 and 1 labels would achieve this.

Other recently added documentation can be found on the cleaver downloads page.
http://www.sci.utah.edu/download/cleaver

We'll also update the documentation to make this single-input-file issue more
clear for the future. And please let us know if you any more trouble with the
software. Good luck.

Cheers,
        Jonathan  Bronson

On 2/1/2013 11:44 PM, Fernando Quivira wrote:

Dear Cleaver developers:

My name is Fernando Quivira. I work with Dana in the Breast Imaging Research Lab at the Mass. General Hospital. I don't know how much has Dana told you about what we do, but we basically need meshing software to create high quality meshes of breast tomosynthesis data for our finite element simulations.

I started using Cleaver with simple datasets, but I have encountered a couple of problems. Here is the info I have compiled for you

System details:

Ubuntu 12.04 - 64 bits
Nvidia GTX 480
GCC 4.6.3 (program did not work with 4.4)
48 GB RAM
(let me know if you need more system info)

Installation:

Everything went smoothly

Usage:

I tried writing NRRD files from Matlab. I found a .m file in the Slicer plugin website, but it did not work. Do you guys have Matlab tools to write NRRD files?
Then, I wrote a simple NRRD writer. However, cleaver was stuck while reading the file. The file used a simple header:

NRRD0001
type: int8
dimension: 3
sizes: 441 280 95
encoding: raw

My next step was to save my data as a .mat file, and use Seg3D to convert it to nrrd. This seemed to work as Cleaver did not get stuck in the reading file process.
Doing this with Seg3D was somewhat difficult since it kept crashing with a seg fault everytime I tried to write data as NRRD. Only occasionally was I able to do it.

Cleaver finished executing with the NRRD data from Seg3D, but the matlab output is empty. Cleaver log shows that the mesh contains 0 elements.
The command I ran was:

cleaver -i ./test.nrrd -o test2 -f matlab

I have attached the data and results as a .rar file.

test.mat contains the simple block to be meshed. test.nrrd is the NRRD file from Seg3D. the files test2.* are the outputs from cleaver
Here is the output from the command line:

Reading File: ./test.nrrd
Input Dimensions: 441 x 280 x 95
Creating Mesh with Volume Size [441, 280, 95]
Cuts Computed
Triples Computed
Quads Computed
Tets Generalized
Phase 1 Complete
Phase 2 Complete
Phase 3 Complete
Stenciling Complete
Mesh Construction Complete
Total Time: 0 s
Worst Angles:
min: 180
max: 0
Writing settings file: test2.info
Writing mesh matlab file: test2.mat
Writing mesh ply file: test2.ply
Cleaning up.
Done.

I don't know what I'm doing wrong. Let me know if there is more information you need. Is there some documentation I can read regarding cleaver usage?

thanks! I hope this is of help

- Fernando