point games
b...@coolgroups.com
where the primitive is points rather than triangles.
Nathan Mates
<b...@coolgroups.com> wrote:
>I was just wondering if we're going to start seeing computer games
>where the primitive is points rather than triangles.
You can do single-pixel points just fine in DirectX right now. [I
assume OpenGL is similar.] However, a huge screen (e.g. 1600x1200 or
bigger) would need a lot of single points to be visible. There's also
the problem that the hardware (graphics cards) tend to support
triangles *great* and everything else not so well. This is a "network
effect" (see http://en.wikipedia.org/wiki/Network_effect ) where
what's supported well gets a lot more support, and everything else
gets ignored.
I suppose the bigger question is *why* would you think that points
(maybe you mean spheres?) would look better?
Nathan Mates
--
<*> Nathan Mates - personal webpage http://www.visi.com/~nathan/
# Programmer at Pandemic Studios -- http://www.pandemicstudios.com/
# NOT speaking for Pandemic Studios. "Care not what the neighbors
# think. What are the facts, and to how many decimal places?" -R.A. Heinlein
John Nagle
> In article <785160de-8577-4950...@s12g2000prg.googlegroups.com>,
> <b...@coolgroups.com> wrote:
>> I was just wondering if we're going to start seeing computer games
>> where the primitive is points rather than triangles.
>
> You can do single-pixel points just fine in DirectX right now. [I
> assume OpenGL is similar.] However, a huge screen (e.g. 1600x1200 or
> bigger) would need a lot of single points to be visible. There's also
> the problem that the hardware (graphics cards) tend to support
> triangles *great* and everything else not so well. This is a "network
> effect" (see http://en.wikipedia.org/wiki/Network_effect ) where
> what's supported well gets a lot more support, and everything else
> gets ignored.
>
> I suppose the bigger question is *why* would you think that points
> (maybe you mean spheres?) would look better?
He might be thinking of a ray-tracer. We still can't quite do ray-tracing
in real time for non-trivial scenes on consumer grade hardware. It's getting
close, though. See "http://www.youtube.com/watch?v=oLte5f34ya8".
John Nagle
Nathan Mates
John Nagle <na...@animats.com> wrote:
>> I suppose the bigger question is *why* would you think that points
>> (maybe you mean spheres?) would look better?
> He might be thinking of a ray-tracer. We still can't quite do
>ray-tracing in real time for non-trivial scenes on consumer grade
>hardware. It's getting close, though. See
>"http://www.youtube.com/watch?v=oLte5f34ya8".
It's been "close" for quite some time. I remember hearing things
like that for at least a decade now. I'm skeptical on it, though I
think that 'real AI' is definitely further off than realtime
raytracing. While CPUs are getting faster and able to crunch more
data, I see two big problems:
(1) GPUs are a form of asymmetrical multiprocessing, offloading at
least some work from your CPU(s). Moving to realtime raytracing will
shift the bulk of the work back onto the CPU, where the CPU is needed
for things like AI, physics, gameplay logic, etc. Just about every
system has a large number of transistors that can push triangles (and
not much else), so if you leave that idle, it'll hurt overall
performance.
(2) Monitors are getting larger. I have a 1920x1200 LCD monitor
(effectively 1080P HD, I believe) on my main box at home. While CRTs
can scale up a lot of resolutions for free, LCDs have a very limited
set of resolutions that look good when scaled to the native
resolution. In the next 2-3 years, I can see more and more consumer's
monitors getting larger faster than CPUs ability to feed them data.
Luc The Perverse
He's probably mistakenly under the assumption that the triangles are
always necessarily bigger than a pixel.
--
LTP
:)
b...@coolgroups.com
> In article <785160de-8577-4950-9424-14e918870...@s12g2000prg.googlegroups.com>,
>
> <b...@coolgroups.com> wrote:
> >I was just wondering if we're going to start seeing computer games
> >where the primitive is points rather than triangles.
>
> You can do single-pixel points just fine in DirectX right now. [I
> assume OpenGL is similar.] However, a huge screen (e.g. 1600x1200 or
> bigger) would need a lot of single points to be visible. There's also
> the problem that the hardware (graphics cards) tend to support
> triangles *great* and everything else not so well. This is a "network
> what's supported well gets a lot more support, and everything else
> gets ignored.
>
> I suppose the bigger question is *why* would you think that points
> (maybe you mean spheres?) would look better?
>
> Nathan Mates
> --
> # Programmer at Pandemic Studios --http://www.pandemicstudios.com/
> # NOT speaking for Pandemic Studios. "Care not what the neighbors
> # think. What are the facts, and to how many decimal places?" -R.A. Heinlein
Points are just easier to work with. You don't have to worry about
the complexities of closing meshes and such. You could probably make
a decent point model just from a 2d image of someone. Doing that
with
meshes is a hassle.
Also, 3d laser scanning translates more naturally into point clouds.
Plus, points are closer to how it is in the real world.
Luc The Perverse
But . . . what happens if you zoom in? Either your "point" has width to
it in which case it is not a point at all, or eventually the target
becomes see through
--
LTP
:)
b...@coolgroups.com
>
> - Show quoted text -
You would want to choose a point density appropriate for the maximum
zoom level.
RazzleGames
Sounds more like spherical approximation (using spheres instead of
triangles)? You could likely change the detail level based on the
distance, in real time...