[osg-users] osgdem --terrain vs polygon count?
robert.osfield at gmail.com
Mon Sep 29 08:02:30 PDT 2008
The main speed up with using --terrain in the osgdem/vpbmaster build
is that the tile geometry doesn't need to simplified - it can just be
aquired from the source DEM's as grids and then output as grids.
At runtime there will typically many more triangles being rendered
with an --terrain (osgTerrain) scene than one with standard paged
database with osg::Geometry, the delta between the two depends upon
how flat the region is, and the error metric used in the
simplification, and the SampleRatio used in osgTerrain at runtime.
The key thing to remember with --terrain databases is that since the
triangulation is done dynamically on load you do load balancing - you
can set the Terrain::setSampleRatio() according to the hardware that
you have, so low level hardware you can set a very low ratio to cull
lots of geometry, while high end hardware may even be able to handle a
ratio of 1.0.
You can't do this type of geometry complexity load balancing with a
polygonal paged database, the best you can do is change the Camera's
LODScale which affects LOD selection and hence which Nodes, Textures
and Geometry that will be selected.
On Mon, Sep 29, 2008 at 3:50 PM, Jean-Sébastien Guay
<jean-sebastien.guay at cm-labs.com> wrote:
> Hi all,
> I admit I don't know what causes osgdem --terrain to take so little time to
> generate the same terrain that would take hours without that option. I
> suspect that's part of the answer to my question. But I'd like to know a few
> of the details if possible.
> I was wondering why, when I generate a database with the --terrain option,
> the geometry is a uniform grid (not simplified)? I would suspect this would
> affect the run time speed on large terrain and where the rest of the scene
> (other than the terrain) is complex?
> So are we trading a (very large) increase in build speed for a slowdown in
> run time speed on slower video cards? What is the magnitude of that slowdown
> in practice (assuming it actually exists)?
> Jean-Sebastien Guay jean-sebastien.guay at cm-labs.com
> osg-users mailing list
> osg-users at lists.openscenegraph.org
More information about the osg-users