[Neurodebian-devel] RE : Packaging of anatomist

[RIVIERE Denis]

Hi Michael,

It doesn't really surprise me that anatomist crashes when its data files are not found: as these files are part of the same project sources, they are not really expected to be missing, and as you have seen Anatomist is a quite big project which has grown with years through different ages, and is not a model of software engineering at every place (we tried to, but...). The crash should not be difficult to avoid, and a clear error message could be issued instead (I have just assigned myself a bug report).

I haven't completely understood what you tried to do: what do you call a "painted volume" ?
Displaying several volumes need to make a "3D fusion" object from the different volumes (using the "fusion" button mechanism): it is a bit surprising at first to require such a user interaction, but then it allows to choose how the different volumes are merged in the interaction. And then you can build other objects from this 2D fusion (like volume rendering, or arbitrarily oriented slices), or involving also other objects (texturing a mesh, or a plane in a cut mesh, and so on). Once you get a bit used to it, the manipulations in anatomist are all similar.

You also have a simpler application based on Anatomist library (and its python bindings): it's called "anasimpleviewer.py" (it's also in the bin/ directory) for more conventional manipulations and a more classical user interface: it performs this kind of fusions automatically as you load volumes for instance (but is fairly limited compared to the full featured anatomist).

Denis

-------- Message d'origine--------
De: Michael Hanke [mailto:michael.hanke at gmail.com]
Date: mer. 09/02/2011 15:22
À: RIVIERE Denis
Cc: Yann Cointepas; neurodebian-devel at lists.alioth.debian.org
Objet : Re: [Neurodebian-devel] Packaging of anatomist
 
Hey,

thanks for all the guidance so far!

On Tue, Feb 08, 2011 at 05:11:02PM +0100, Denis Rivière wrote:
> we always use an environment variable ($BRAINVISA_SHARE) pointing to the share/
> directory. Normally if it is not present, it defaults to a "more
> normal" path, but I have to check whether it defaults to
> /usr/local/share or to /usr/share, or relatively to the executable,
> or looks to sevral of them. I will look at it and tell you. Maybe
> it's the key of the problem.

Yes, that was the key. Maybe I was a little confused that missing
datafiles would result in a segfault ;-) Setting the BRAINVISA_SHARE
variable to the proper location allowed me to start anatomist.

It will take me a while to familiarize myself with anatomist. My initial
attempt to put a 'paint volume' on top of a volume rendered anatomical
failed badly -- it looks like a complex piece of software ;-)

I'll be on vacation next week, and will continue with this project
afterwards. I started taking notes on some issues that we should discuss
in terms of optimal packaging of brainvisa (starting with anatomist, but
with the rest already in mind).


Cheers,

Michael

-- 
Michael Hanke
http://mih.voxindeserto.de