[Bf-committers] Bachelor Thesis in Parametrization of Meshes

[Brecht Van Lommel]

Hi Aurel,

The results looks great! You can mail me with questions, or better,
create a task or code review on developer.blender.org, and then we can
discuss things there and it's all logged publicly for anyone in
interested now or later.

Regarding where this code fits, there's a few possibilities:
* Write algorithm directly integrated in Blender mesh code, using the
BMesh data structure
* Create a new module in the intern/ directory, using own data
structures, code is maintained by Blender developers
* Add libraries in the extern/ directory, code is not maintained by
Blender developers but bundled with the source code
* Use an external library, code is not maintained by Blender developers

There's some trade-offs here. External libraries make building and
packaging Blender harder, and it requires the developers of that
library to maintain and support it on all our platforms.  For a single
mesh tool like this it would be best if the source code was bundled
with Blender I think, often such libraries tend to not to be
maintained for long and Blender developers have to start fixing things
themselves anyway. If it's only a couple thousand lines of code total,
rewriting it using Blender data structures may be best in the long
term. We'd need more details to decide on that.

Brecht.


On Thu, Mar 24, 2016 at 4:16 PM, Gruber  Aurel <augruber at student.ethz.ch> wrote:
> Hey Guys,
>
> I’ve been using blender for several years and am now a Computer Science student at ETH Zürich and currently doing my Bachelor Thesis. It’s about large scale mesh parametrization.
>
> The IGL<http://igl.ethz.ch> department at ETH developed a new algorithm that, altough not quite as fast as the rather simple LSCM, delivers much better results and minimal stretch. You can find a blender file with a comparison with two uv maps here<https://drive.google.com/folderview?id=0B8tVvh1f2GtPemVKNFY1Si0zS2M&usp=sharing>. The paper has not yet been published which is why I can’t send you a link to that, but if someone is interested i can do so as soon as it gets published.
>
> My Thesis is now about implementing that in blender, optimizing some parts of it (such as choice of solver, and acceleration with momentum functions) and evaluating the results. I have written some python plugins for blender in the past for personal use but haven’t messed with the blender source so far. I would therefore be very grateful if I may ask some questions from time to time and it would be awesome if you could direct me to the right people. From what I’ve seen in the dev wiki, Brecht worked on UV unwrapping, right? Can i contact him somehow?
>
> Also, since the algorithm uses some external libraries from IGL and Pardiso (although it falls back to EIGEN if Pardiso is not found) how are such scenarios usually handled? Should everything be rewritten to comply to Blender code conventions? Can libraries be used if the license permits it?
>
> Thanks to all the people who have worked on Blender so far, it’s truly awesome and I’m glad i get a chance to work of some of it too!
>
> Aurel
>
>
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers