[osg-users] Searcing for raytracer implementation based on OSG

Adrian Egli 3dhelp at gmail.com
Sun Oct 14 11:53:35 PDT 2007


Hi all,

i did a really first implementation of RayTracing with OSG to improve how
fast my kdTree intersection test is. Check out the kdTree.tar and test it.
(Are there someone who has time to integrate it into OSG core for boosting
the Intersection Visitor ? )

/adegli

KD-Tree...[23348 Triangles:20 rays]
KD-Tree Build Time      =       0.224797s
******** TEST 1 **** (KDTree / OSG Intersection Test)
Intersection Time       =       0.000168s
Rays                    =       1
Nodes Checks            =       79
Bounding Checks         =       122
Triangles Checks        =       49
Linehits                =       6
Intersection Time <osg> =       0.018305s
Linehits                =       6

SpeedUp                 =       108.958 (FACTOR !, 20 rays)

******* TEST 2 *****
Ray Trace Time          =       2.97638s

Rays                    =       787307          (264K Rays/s)
Nodes Checks            =       24939199
Bounding Checks         =       49512447
Triangles Checks        =       16703960

Viewer...

2007/10/11, Adrian Egli <3dhelp at gmail.com>:
>
> Hi Joachim,
>
> isn't at all a stupid question, of course i have to care about this, but i
> still thought about this issue,
> may there a two possiblity, first i store the data still in a array and
> just refere to them from kd-nodes,
> or a link the kd-nodes, i we assume that only triangles are stored in the
> leaves, then may we link juste the leaves.
>
> (a stupid idea: may we can "link them to a ring", then may we can order
> them in depth, may not. i have to think about this,
> but i we have a strategy to walk through depth sorted you will see a next
> application of kd-tree (heuristically depth sorted) )
>
> /adegli
>
> 2007/10/11, Joachim E. Vollrath < vollrath at vis.uni-stuttgart.de>:
> >
> > Hi Adrian,
> >
> > possibly a stupid question:
> >
> > if you do that, wouldn't you break the vertex arrays that are needed to
> > efficiently upload the geometry to the graphics card?
> >
> >
> > > Hi Joachim,
> > >
> > > at the moment it's only a quick and dirty evaluation prototypic
> > > implementation, of course i have a huge overhead, i like to replace
> > for
> > > triangles
> > > mesh the vertex array by a kd-tree in the osg core implementation. may
> > i get
> > > time tomorrow to doing depther tests and some may i will be able
> > > work out a first version of such a replacement in the osg core.
> > >
> > > i guess if i will integrate it right in osg core  the overhead will be
> > not
> > > really huge, but i don't yet know it exactly
> > >
> > > /adegli
> > >
> >
> >
> > --
> > Joachim E. Vollrath
> > University of Stuttgart
> > Institute for Visualization and Interactive Systems (VIS)
> > web:   http://www.vis.uni-stuttgart.de/~vollrajm/<http://www.vis.uni-stuttgart.de/%7Evollrajm/>
> > mail:  vollrath at vis.uni-stuttgart.de
> > phone: 49.711.7816.403
> > fax:   49.711.7816.340
> >
> > german postal address:
> > Dipl. Inf. Joachim E. Vollrath
> > Universitaet Stuttgart, Institutsverbund Informatik
> > Institut fuer Visualisierung und Interaktive Systeme
> > Universitaetstrasse 38
> > 70569 Stuttgart
> > GERMANY
> >
> > _______________________________________________
> > osg-users mailing list
> > osg-users at lists.openscenegraph.org
> >
> > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
> >
>
>
>
> --
> ********************************************
> Adrian Egli




-- 
********************************************
Adrian Egli
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openscenegraph.org/pipermail/osg-users-openscenegraph.org/attachments/20071014/ef4ef907/attachment-0003.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: kdTree3.tar
Type: application/x-tar
Size: 51200 bytes
Desc: not available
URL: <http://lists.openscenegraph.org/pipermail/osg-users-openscenegraph.org/attachments/20071014/ef4ef907/attachment-0003.tar>


More information about the osg-users mailing list