-----Original Message-----
From: Akbar A. [mailto:syedali011@earthlink.net]
Sent: Wednesday, December 13, 2000 6:49 AM
To: gdalgorithms-list@lists.sourceforge.net
Subject: RE: [Algorithms] Simple 3D engine design.


>Once DX8 hardware becomes available
>Why bother doing this per-pixel? Because it looks great!
:)

yeah.

but it's been mapped to hardware for a while now with nvidia hardware
and how many games are using it?

i dunno of any, but i haven't played to many games in a while.
i'm thinking the crytek engine, and maybe croteam...

same with ati's stuff.
you'd think more people in there commercial games would be scrambling to get these features supported in there games.

but usually when i talk to those people they give me the
"we would concentrate more on the game side".
"there just not enough people with geforce's for me to use those extensions you say are cool"

i understand first part, 
but the second excuse is kinda of sad.
seeing that a "conservative average" of nvidia's line is say, 100 US (geforce and tnt2) ?

then take the amount of money nvidia generated for fiscal year 2000 iirc
it's at 700 million US $.

if you do the math (even with dev/production costs), that's a hell of a lot of video cards that are using the geforce set.
i'm not sure what was there big seller for this year, but i'm guessing the geforce chipset.

and the 'features' ati's and nvidia's extended opengl come with, if you don't support them you loose a "BIG" boost in scene quality.

but, i think with the geforce-mx nvidia' kinda of scimped out on the pixel pipeline?
if they did i don't blame them. 
are there even any games out yet that are pushing(need the clock speed) the hardware (texturing or geometry wise)?
memory fading me...

so if your afraid that your code passes won't work on many of your customers, 
just know that there are _tons_ of geforces out there and they are craving to have there fragment colors calculated through the register combiner pipeline :)

i'm thinking that was a tad bit to organic for some hardware, break out the nv20 :)

pixel shaders have there drawbacks, i wrote but i deleted it :|

passing out at 6:48 am :|
laterz,
akbar A.


;vertexabuse.cjb.net


"necessity is the mother of strange bedfellows"
-ninja scroll

-----Original Message-----
From: gdalgorithms-list-admin@lists.sourceforge.net
[mailto:gdalgorithms-list-admin@lists.sourceforge.net]On Behalf Of Alex
Clarke
Sent: Wednesday, December 13, 2000 3:10 AM
To: 'gdalgorithms-list@lists.sourceforge.net'
Subject: RE: [Algorithms] Simple 3D engine design.


Interesting idea, one slight problem :-)

Once DX8 hardware becomes available, I think per-pixel lighting is
going to be come ubiquitous.  Of course if people use lighting with
shadows, there is a scaling problem:
		
colour = SUM(Diffuse(i) * Shadow(i)) * texture + SUM(Specular(i) *
Shadow(i))

Unless you have two render contexts or two accumulation buffers, you
are going to need an awful lot of passes to render half a dozen lights...

Why bother doing this per-pixel? Because it looks great!
   
I think we can find a use for as much extra performance
as the card manufactures can possibly give us...

Alex Clarke, Programmer
Argonaut Games PLC

------------------------------------------------------------------------
Thatcher Ulrich said:
[1] So does this mean that there's no point in GPUs going facter than
a certain tris/sec level (somewhere between 100-200 Mtris/sec)?  I
think so, at least not until we're targeting super hi-res displays.
The looming problem is how to come up with enough worthwhile triangles
to cover the screen with interesting detail.  Perspective projection,
interactivity and large view distances dictate extreme detail
scalability.  As a sort of rough limit estimate, lets say you have a
90-degree FOV, you want 1-pixel features, your near clip is 0.1 meter,
and the player can go anywhere in a 10km x 10km map.  Then your world
must contain say 10^16 triangles.  Those requirements may be a bit
extreme, but I think that's where we're headed next.
_______________________________________________
GDAlgorithms-list mailing list
GDAlgorithms-list@lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/gdalgorithms-list