DiagSplit

[Anteru]

Hi,

the DiagSplit paper has been released:

http://graphics.stanford.edu/papers/diagsplit/diagSplitPaper.pdf

Cheers,
  Anteru

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
Aqsis-development mailing list
Aqsis-development@...
https://lists.sourceforge.net/lists/listinfo/aqsis-development

[Chris Foster-5]

On Wed, Sep 16, 2009 at 6:28 AM, Anteru <Anteru@...> wrote:
> Hi,
>
> the DiagSplit paper has been released:
>
> http://graphics.stanford.edu/papers/diagsplit/diagSplitPaper.pdf

Hah, what do you know, they referenced us (hmm, [Foster2009]
apparently :-) ) for a crack avoidance algorithm, which is nice I
guess, though I'm certainly not the only one who contributed to the
wiki page of potential solutions (
http://wiki.aqsis.org/dev/grid_crack ).  I am a bit bemused however
given that we haven't actually implemented any of the crack avoidance
algorithms in aqsis itself, and therefore we don't know if the ideas
which are on the wiki are any good at all.

I've thought about the issue on and off for a while now, always
concluding that inter-patch crack stitching *requires* information
from both neighbouring patches.  As far as I can tell, *general*
displacement shading on a grid cannot reproduce the same displaced
positions on neighbouring patch edges, with the prime culprit being
derivatives: derivatives must be one-sided at grid boundaries, and
they depend intimately on the dice resolution.

As I see it, the only choices are:
1) Give in, and make use of grid adjacency information
2) Somehow do significant redundant shading work at patch edges to
carefully reconstruct the vertices of neighbouring patches without
needing to store and look up stitching data.  This approach would be
parrallelisable but fraught with difficultly even if the derivative
problem could be solved (eg, what about facevarying values which are
fundamentally discontinuous?).

Anyway, thanks Anteru for linking this paper for us to have a look at!
 AFAICT from a quick skim they make specific assumptions about the
nature of the displacement shading which would be a good trade off for
realtime graphics, but not helpful to solve the cracking problem in
the presence of general displacement shaders.  Hopefully I'm wrong!

~Chris

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
Aqsis-development mailing list
Aqsis-development@...
https://lists.sourceforge.net/lists/listinfo/aqsis-development