Invisible wrote:
> POV-Ray, of course, gets round this problem by using more sophisticated 
> mathematical techniques than simply projecting flat polygons onto a 2D 
> framebuffer. I've yet to see any GPU attempt this.

Well, I wrote a vertex shader a while ago that would correctly position 
a point along a Bezier patch*.  Then, rendering the patch became a 
matter of passing a "net" of points in, and let the GPU handle the 
actual calculation of the points.

It wouldn't be too much of a stretch to extend this to other functions, 
and not just a Bezier calculator.

-- 
...Ben Chambers
www.pacificwebguy.com

*It even varied the granularity of the net in real time based on the 
current FPS, so it would maintain a constant frame rate.  It ended up 
quite nice, and I was able to have about a dozen patches on screen, with 
no visible polygon edges, on my GeForce 5200.

Of course, on my 7600GT, I can render several dozen patches on screen at 
once, with no visible polygon edges.  Still a far cry from modern games 
(where an individual model might have 8-15 *thousand* polygons, and 
there are usually several characters on screen at once), but a neat 
effect anyway :)

Post a reply to this message

Invisible wrote:
> Also, how does it figure out which polygons to draw? It can't possibly 
> draw all 10 million polygons every frame - and yet, figuring out which 
> ones are visible would seem to take more effort than actually drawing 
> them all...

Something similar to bounding volumes (though they're really visibility 
volumes), to limit the areas that are drawn.

For instance, when you're inside a room, the engine knows that it 
doesn't need to draw the rest of the map - it only draws that room. 
Furthermore, it also knows precisely which areas are visible outside the 
window, and which ones aren't.

These volumes are usually computed ahead of time, as a large map can 
take up to several minutes to compute the volumes (depending on the 
game, the level, and the machine being used to build it).

-- 
...Ben Chambers
www.pacificwebguy.com

Post a reply to this message

Invisible wrote:
> I often look at a game like HL and wonder how it's even possible. I 
> mean, you walk through the map for, like, 20 minutes before you get to 
> the other end. The total polygon count must be spine-tinglingly huge. 

It doesn't draw 10 billion separate vertices - it only draws about 150M 
vertices, 60 times each.  Of course, most of those are instances of the 
same meshes, so it's really only about 1-3M unique vertices in a given 
scene.

-- 
...Ben Chambers
www.pacificwebguy.com

Post a reply to this message

nemesis wrote:
> BTW, is it just me or anyone else feel like it's easier to add new code and
> lighting models into povray rather than writing a povray SDL parser to other
> renderers? :P

Wasn't that one of the original design goals? ;)

-- 
...Ben Chambers
www.pacificwebguy.com

Post a reply to this message

Chambers wrote:
> Until it can parse SDL, I will consider it inferior to POV.
> 
> Long live hand written code! ;)

I think I just found the user requirements spec for my new raytracer...







;-)

No, but seriously, there are some things about SDL that I'd change. Like 
making it possible to inspect existing objects and edit their textures 
and stuff...

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message