Re: [brlcad-devel] Introduction for GSoC 2012.

Wed, 14 Mar 2012 07:45:19 -0700

n Mar 14, 2012, at 05:45 AM, Animesh Garg <garg.anim...@gmail.com> wrote:

Hello BRL Developers,
 
Hello Animesh!  That's a great 3D-graphics name you have there. ;)

I have a background in optimization & stochastic modelling (operations research), robotics, computer vision. I have taken graduate courses in vision, studied and taught undergraduate Computer Graphics and Database classes. I have sufficient programming  experience, mainly in c/c++ and matlab for robotics/vision domains where I implemented custom algorithms and worked with robot hardware. Furthermore, I am comfortable using either Linux or Windows. I have worked on projects with subversion.

Certainly a lot of BRL-CAD relevant background experience you have there, welcome to the community.  Most of our projects that will probably be most interesting to you are C/C++ so it sounds like a good fit.

I have been glossing over project ideas. Most of the development at brl-cad seems to be in c/c++ barring a few projects.
Top projects that interest in me , in order of preference, are :

We're adding new ideas every day so don't be shy to suggest a new idea as well if there's something in particular that piques your interest we haven't mentioned.  Particularly with your background in computer vision, there's a lot our ideas page doesn't mention but BRL-CAD has been involved with in the past. 

 
How good are you reading existing code?  What about somewhat complex C++ template class code?  Our libpc parametric constraint library is built on top of Boost's Spirit library.  Dawn can certainly speak more to the current status but I can say that this is a high-priority hot topic, so glad to see you're interested.  As it is, the library is in place and it has test code that evaluates basic expressions correctly.  We're actually in the process of integrating the existing solver in with our primitives right now so that known implicit constraints can be automatically evaluated on-demand.  From there, the intent is to have explicit constraint entities that describe one or more relationships on geometry.


This is a fun project that could either be implemented in the output 2D image-space domain or as a projection of 3D geometry.  The prior is sampled, the latter is not.  The latter is the more desirable way to go, but the one where we lack a critical routine (evaluate surface-surface intersections of NURBS surfaces).  Doing it in image-space should be suitable but begs for adaptive sampling to make sure detail is captured appropriately.


This is a cool one.  It's more about API restructuring than anything, but the impact potential is pretty grand (most CAD systems can't do this).  The idea is to parametrically describe an object's material properties as a set of density functions.  Given a particular ray slicing through an object, you'd get back a density function for all segments hit by the ray.  An easy way to conceptualize this is to take a "core sample" of a bone.  Bone is considerably more dense near the surface and gets progressively less as you move towards the center.  You'd probably define density as a function of distance from the surface or distance from the medial axis.

I would be glad to discuss possible proposal with brl mentors before finalising my idea. It would be great to get input on current state of the above problems, and also what would be a good set of goals, non-trivial but not impossible, to complete during the summer.
 
Follow-up questions are certainly welcome (and other devs are encouraged to chime in if I left anything out or you just want to add something to the discussion).

Cheers!
Sean

------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/
_______________________________________________
BRL-CAD Developer mailing list
brlcad-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/brlcad-devel

Reply via email to