Thank you for answering my question in such detail. I've been looking
through the documentation and source code, but I was having a bit of trouble
figuring it all out, so your reply was very helpful and I'm glad to see that
someone might be interested in this idea.
On 21 March 2010 18:10, res <[email protected]> wrote:
> On 18.03.2010 21:08, Mike Gist wrote:
> > We currently support shadow mapping, however plenty of work can be done
> > on it to improve it. Frank would know more about that though.
>
> In principle, it's there. But it isn't used too much; also the
> implementation is _not_ pretty.
>
> One thing I believe that would both improve shadows + make the
> implementation nicer is to use VSDCT for point lights.
> That letter monster unfolds to "Virtual Shadow Depth Cubemap Texture"
> (unless I swapped some letters...). It's a solution for the problem that
> you don't have "depth cubemaps" on all hardware (afaik, DX10-class HW
> has it, DX9 not); so what you do is to render the six cubemap faces to
> one 2D depth texture, at different places.
> Compare to what CS does now for point lights - 6 shadow maps, one for
> each side! - and you see why this is desireable...
> Spot lights, IIRC, are not much better, taking up to 5 maps (if the spot
> angle is >90°).
>
> Soft shadows: there's currently some support for VSM, but even less used
> than regular shadow maps. But last time I checked, that worked, and
> certainly something to start on.
>
> I think a shadows GSoC project could look something like this:
> 1. Check state of shadows for directional lights, make sure they work.
> (These are the easiest shadows.)
> 2. (Maybe) check state of shadows for "tight" (<= 90° angle) spotlights
> (they're still easy, but a bit harder than directional light shadows.)
> 3. Implement VSDCT for "wide" spot lights + point lights. (That's the
> kicker.)
> Beyond that, all optionally:
> 4. Make sure there are some nice demos - eg 'castle' with realtime
> shadows, isotest w/ shadows, add a "flashlight" to walktest to show off
> realtime shadows ...
> 5. Look into soft shadows - eg make sure CS' VSM works. Possibly look
> into other methods.
>
> -f.r.
>
>
>
> ------------------------------------------------------------------------------
> 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
> _______________________________________________
> Crystal-main mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/crystal-main
> Unsubscribe: mailto:[email protected]
> ?subject=unsubscribe
>
>
------------------------------------------------------------------------------
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
_______________________________________________
Crystal-main mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/crystal-main
Unsubscribe:
mailto:[email protected]?subject=unsubscribe
