Re: [Algorithms] Silhouette determination for the world's shadow volume
Brought to you by:
vexxed72
Menu
▾
▴
Re: [Algorithms] Silhouette determination for the world's shadow volume
|
From: Matt J <mjo...@gm...> - 2010-04-18 21:59:17
|
A lot seem to forget finding edges on the CPU is 'optional'...and time consuming. So you have two options here..you could do the edge finding using a geometry shader. Another approach that works on earlier hardware is have a shadow mesh, simplified, and independent of the primary mesh. You can duplicate each edge (with opposite winding, and connected as quad to the first edge) except store the coordinates of the extra edge with a 0 w coordinate... and then all you have to do in the vertex shader is extrude out the edge with 0 w coordinate to infinity. At the cost of extra geometry (which can be simplified independently), you keep the processing on the GPU. This works fine as long as the geometry is solid. It doesn't need to be convex, nor really solid either (although you really want too, to avoid artifacts). You're extruding it as a preprocessing step, at the expense of greater geometry. However, you can use a simplified mesh so you just keep the core 'features' of the mesh, e.g. you can use a few AABB to represent a building mesh.Finally, you can hide the shadow artifacts on the backside using reverse extrusion. With modern shadow algorithms though you are going to get much nicer shadows using a shadow buffer. Thanks, Matthew Well, the shadow volume mesh topology rules really aren't any different for > light sources inside a mesh vs outside - as long as your geometry faces > inwards (and it will), the idea is the same (find and extrude edges that > border lit & unlit faces). Sometimes world geometry is stored in funny ways > depending on your game/engine (BSP trees or what have you), that might make > finding edges trickier, but the algorithm is the same. > > A good optimization for fill rate if you are using local light sources is > to make sure and only extrude shadow volumes for the areas of the world that > intersect the light's attenuation radius. > > Jeff > > > On Sun, Apr 18, 2010 at 3:12 PM, Damian Coventry < > Dam...@m8...> wrote: > >> Morning, >> >> WRT shadow volumes, what's a common approach for determining the >> silhouette for the world mesh? Let's say it's an indoor only game. In >> this case the camera and lights are both "inside" the world mesh, and >> the world mesh is concave. >> >> For the other meshes that move around inside the world, silhouette >> determination is well documented - "The set of all edges that connect a >> toward-face to an away-face form the silhouette with respect to the >> light source. The edges forming the silhouette are extruded away from >> the light to construct the faces of the shadow volume". >> >> What about the world mesh though? My initial thoughts are to use all >> world mesh polygons that don't face the light source, and create >> individual shadow volumes for each polygon. This would get costly if >> the world mesh had many back facing polygons at any given time. How >> else is it done? >> >> >> ~Damian(); >> >> >> >> ------------------------------------------------------------------------------ >> Download Intel® Parallel Studio Eval >> Try the new software tools for yourself. Speed compiling, find bugs >> proactively, and fine-tune applications for parallel performance. >> See why Intel Parallel Studio got high marks during beta. >> http://p.sf.net/sfu/intel-sw-dev >> _______________________________________________ >> GDAlgorithms-list mailing list >> GDA...@li... >> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list >> Archives: >> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list >> > > > > -- > Jeff Russell > Engineer, 8monkey Labs > www.8monkeylabs.com > > > ------------------------------------------------------------------------------ > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > _______________________________________________ > GDAlgorithms-list mailing list > GDA...@li... > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > Archives: > http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list > -- ----- Matt Johnson http://otowngraphics.blogspot.com |
