[Soc-2010-dev] Sculpt Brushes Status Report - 5/28/2010

Jason Wilkins jason.a.wilkins at gmail.com
Sat May 29 02:02:41 CEST 2010


If '|' is the flatten plane, and > and < represent the attractive force it
has on vertexes, and left is above the plane and right is below the plane
then:

Flatten:       > | <
Contrast:     < | >
Fill:                | <
Scrap:         > |

Scrap very convincingly and naturally looks like scraping off clay if you
have peeks and valleys.  Its kinda scary.
Fill does the opposite, it fills in the valleys and leaves the peaks alone.

On Fri, May 28, 2010 at 5:43 PM, Nicholas Bishop
<nicholasbishop at gmail.com>wrote:

> Nice work here. I particularly like your changes to brush spacing,
> using the brush size instead of pixels makes so much sense in
> retrospect, good catch there :)
>
> I'm curious if you could describe more what the differences between
> the flatten, fill, and scrap brushes are?
>
> Thanks,
> -Nicholas
>
> On Fri, May 28, 2010 at 7:23 AM, Jason Wilkins
> <jason.a.wilkins at gmail.com> wrote:
> > I focused on three areas this week:
> > 1.) Plane brushes - These are brushes that all determine the average
> plane
> > of the vertexes under the brush and then pull or push vertexes relative
> to
> > the plane.  These include
> > - Flatten - pull vertexes towards the plane
> > - Clay - offset the plane above or below the surface then draw vertexes
> > towards it
> > - Fill - pull vertexes up towards the plane
> > - Scrap - pull vertexes down towards the plane
> > - Contrast - push vertexes away from the plane
> > 2.) Optimization - I've used a profiler to find functions that are taking
> > large amounts of time and I optimize those functions (Wash, Rinse,
> Repeat)
> > - eliminated dead code
> > - removed unnecessary code
> > - simplified overly complicated code
> > - removed calls to calloc and memset
> > - created multiple versions of a function when the default does too much
> > work in some cases
> > - however, I have not done any work to improve algorithms
> > - I removed OpenMP for now, pending a better understanding of how to use
> it
> > effectively
> > - for smaller models (1 million polygons) the bottlenecks are no longer
> in
> > the sculpt code itself, but elsewhere
> > - for large models (6 million polygons before my machine croaks),
> optimizing
> > sculpt more would help
> > 3.) Brush Spacing - The distance between stamps is now based on the size
> of
> > the brush instead of screen space pixels
> > - this improves the consistency of spacing under different perspectives
> and
> > brush sizes
> > - the strength of the brush is attenuated according to how much stamps
> > overlap
> > I'll be taking it easy this weekend starting Friday.  I'll be monitoring
> > feedback and planning my next move.
> > Later,
> > Jason Wilkins
> >
> > _______________________________________________
> > Soc-2010-dev mailing list
> > Soc-2010-dev at blender.org
> > http://lists.blender.org/mailman/listinfo/soc-2010-dev
> >
> >
> _______________________________________________
> Soc-2010-dev mailing list
> Soc-2010-dev at blender.org
> http://lists.blender.org/mailman/listinfo/soc-2010-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/soc-2010-dev/attachments/20100528/826f9371/attachment.htm 


More information about the Soc-2010-dev mailing list