[Soc-2014-dev] Weekly Report #1 Shape Key Enhancements

[Sergey Sharybin]

Hi,

We can not assume that if number of verts/edges/faces is the same topology
is the same. Simple example: edge rotate operator -- it changes topology
but keeps the same number of geometry.


On Fri, May 23, 2014 at 4:13 PM, Grigory Revzin <revzingg at gmail.com> wrote:

> Hello everyone!
>
> This (and previous three!) week's progress:
> - Got a good wiki page on  what I'm going to do, including the Midterm
> Evaluation
> - Got git access form Sergey, thanks.
> - Learned quite a few of Blender code.
> - Started the hardest part of refactoring some editmesh conversions
> code. - Implemented rought support for the scratch shapekey lready
> have a rough support for the scratch keyblock. Keyblock recalculation,
> switching.
> - Sped up keyblock recalculations by checking for topology changes in
> the EditMesh. If no topology changes have occured, a faster keyblock
> recalculation algorithm is used (assumes the topology is the same).
> Topology checks are implemented via hashing the editmesh's data on
> shapekey switching and editmode enter/exit.
> - Improved the Shape Key Move operator -- now can move to top of the
> list and to the bottom with one click. TODO: do a double-triangle icon
> for that.
>
> What I'm working on now:
> - Getting mixing shapes in editmode to work. My thought is that using
> the shapkey modifier for that is leading to unnecessary complications
> in its code (it needs to be aware of the scratch in edit mode, also,
> it ponentially e), so editmode mixing code is independent from the
> modifier. This includes enabling editmode independent mix and the UI
> for it.
> - Adding more technical details to the wiki before I forget stuff
>
> Next two weeks:
> - Split the wiki page to distinguish between User's Pov, Tech Notes
> and Progress reports.
> - Stabilize everything described above. There're still segfaults from
> time to time. Add the "Commit to Active" and "Update from Active"
> operators as described in the wiki.
> - Update editmesh undo support, now it seems I've broken it somewhat
> with the refactor.
> - Get a testbuild at the end of the second/start of the third week and
> get someone to test it. :)
>
> Problems:
> - One week before the finals! Lost of time is lost on computing axle
> loads and Maxwell-Mohr displacement integrals, as well as common
> emitter amplifier transfer functions and many other things. Hope to
> have mixing working by end of Sunday.
> - Nevertheless, it's the final weeks that are most demanding, but not
> the finals themselves, so it's going to get better.
> - Is it safe to assume that if totvert/totedge/totface/totloop didn't
> change, the topo is the same? Can we drop hashing?
>
> - Thanks,
> Grigory
> --
> Г. Г. Ревзин
> _______________________________________________
> Soc-2014-dev mailing list
> Soc-2014-dev at blender.org
> http://lists.blender.org/mailman/listinfo/soc-2014-dev
>



-- 
With best regards, Sergey Sharybin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/soc-2014-dev/attachments/20140524/93333932/attachment.htm