[Bf-committers] Sculpting proposal for Google SoC

[Brecht Van Lommel]

Hey Nicholas,

Nicholas Bishop wrote:
>  I'm interested in getting input on a Blender project for the Google
> SoC. (Thanks to Tom M for the idea.) Here's the overview:
> 
> * Integrate SharpConstruct's modeling tools into Blender:
> 
>   SharpConstruct has sculpting tools similar to ZBrush, or to use a
> Blender example, the B-Brush script. As the author of SharpConstruct,
> I expect adding the sculpting tools to Blender will be relatively
> easy; the hard part will be familiarizing myself with Blender's
> existing code.
>   This step will involve adding at least displacement and smoothing
> brushes. The displacement brush is the basic tool that adds or
> subtracts depth from the model. The smooth brush works just like
> Blender's smooth tool, except it only affects the area being brushed
> on.
>   Time permitting, SharpConstruct has two other brushes, pinch and
> grab, that could also be implemented in Blender.
>   SharpConstruct's website is http://sharp3d.sourceforge.net

Sounds great. Given that you already have experience with this, it
certainly looks like this could make a very successful SoC project.

> * Multi-Resolution Modeling (MRM) tools:
> 
>   Once the sculpting tools are done, the fun part begins. MRM stores
> each level of subdivision as a level of detail rather than simply
> replacing the model's previous topology. This allows for the following
> work-flow:
> + Sculpt the model at a high level of subdivision
> + Switch to a lower level of subdivision
> + Make a large changes at the low-res level
> + Return to the high-res level, where the high-res details will be
> reintegrated with the modifications from the low-res level
> + Continue sculpting at the high-res level
>   The primary advantages to this technique are that it's easier to
> make large changes to the model even after a lot of small detail
> modeling has been done, and that it's possible to do this with a
> higher level of interactivity with the user, because the computer
> isn't bogged down with a high-res mesh.
>   MRM will be a more difficult task than the sculpting tools, because
> I haven't ever implemented MRM before. However, there are plenty of
> papers available with possible algorithms.

Yes, I think having support for multiresolution is essential for good
integration with Blender. For example, animating with such a high
poly mesh is not very practical.

Personally I think it would be best to have this working together or
integrated with the subsurf modifier. Displacements could be stored
as a layer of data attached to faces, and the subsurf modifier could
then use the displacements as it subdivides the faces. I believe
ZBrush also somehow interleaves subdivision surfaces with the
displacements.

Brecht.