"Wolfgang Draxinger" <wdraxinger@[EMAIL PROTECTED]
> wrote in message
news:3upqn5-rvm.ln1@[EMAIL PROTECTED]
> Gernot Frisch wrote:
>
>> and if your card does not sup****t PBuffers? Radeon Mobility
>> 7500 comes to mind.
>
> Well, there's still the multipass implementation. Render a pass
> for each subset of the bits. If you need depth testing, render
> the first pass with the depth testing settings you need, then
> for the subsequent p***** keep the content of the depth buffer,
> disable depth write and use GL_EQUAL for the depth test. This
> will also early Z cull occluded fragments.
>
> The nice thing is, you can put all this (using PBuffer or not,
> eventually render muliple p***** for the higher bits and so on)
> into one single function. There are only two variables: Is a
> PBuffer avaliabe. And the number of bits that the colour buffer
> can hold.
>
> Would also make a nice template function. In C++, D or a language
> with similair capable templates. Or using C macros.
>
> Wolfgang Draxinger
> --
> E-Mail address works, Jabber: hexarith@[EMAIL PROTECTED]
ICQ: 134682867
>
Gosh guys, this seems like a lot of bother compared to OpenGL selection
which may be plenty fast enough, and vastly less complicated ( hence, less
buggy). OpenGL selection with a narrow pick aperture works well w/nVidia
cards as they can view frustum cull at a fast rate. I've found selection
to
be usable up to millions of triangles.
jbw
|
