[gameprogrammer] Re: culled objects and shadow volumes
- From: Alan Wolfe <alan.wolfe@xxxxxxxxx>
- To: gameprogrammer@xxxxxxxxxxxxx
- Date: Tue, 28 Apr 2009 11:12:04 -0700
Man, you have been a great help, thank you again Sami (:
On Tue, Apr 28, 2009 at 10:55 AM, Sami Näätänen <sn.ml@xxxxxxxxxxxxxxx>wrote:
> On Tuesday 28 April 2009 18:42:11 Alan Wolfe wrote:
> > That is some pretty good advice, thanks Sami.
> >
> > I found this late last night and this technique talks about expanding the
> > view frustum and then doing regular object culling so you don't have to
> > create the shadow volumes or test which direction the shadows will be
> cast.
> >
> > http://www.gamedev.net/columns/hardcore/shadowcast/
> >
> > I'm going to have to look at the pros and cons of your method vs this
> > method and see which makes more sense for my game cause I was thinking
> this
> > article was the best way, but now im not so sure (:
>
> It really depends on the light positions and types of the lights you use.
> If
> all the lights have short intencity fallout radius then adding that maximum
> radius to the frustum is enough and this is easier to the CPU, but it will
> stress the GPU more.
> If the light sources are always quite a bit higer than the objects then
> extending the frustum might be an option regardless of the intencity
> fallout
> radius.
>
> The thoughest cases are the ones where long shadows are generated from the
> objects that are nearly in the same level as the lights. the toughest
> example
> would be sun set and sun rise. Also all kinds of columns are potential long
> shadow producers.
>
>
> Also consider that the idea of checking the smallest distance to the
> frustum
> position on the OL line is not full proof when the OL line is nearly
> parallel
> with the frustum. In the extreme case the OL line can even go away from the
> frustum and still the shadow volume would intersect the frustum. But these
> cases can be handled separately if such cases exist. These come in to play
> when the light source is near an object as this heavily alters the shape of
> the objects shadow volume. This can be solved by simply calculating the
> same
> condition used in the single test individually for the furthest and the
> nearest point in the OL line that has a perpendicular point in the frustum.
> If either test is true then the shadow volume will intersect the frustum.
> The
> nearest position of course can't be closer to the light source L than the
> object O itself.
> Also there might be extreme cases where the whole frustum is inside the
> objects shadow volume. This means simply that we don't use the light source
> for this scene.
>
>
> So it might be better to implement both methods and use one of them
> depending
> on the combination of the light source and the object.
>
> > Thanks!
>
> Clad that I could help you.
>
>
> ---------------------
> To unsubscribe go to http://gameprogrammer.com/mailinglist.html
>
>
>
Other related posts:
