 News | Forum | People | FAQ | Links | Search | Register |
Log in |
|
|
Using a similair lightmap layout to do edge marking for static shadow volumes. Would require compiler support etc...
Probably wouldn't work since you'dd get discontinuities between lightmap and rt shadows due to blending and interpolation limits etc... Otherwise writing light indexes into lightmaps but then you need to worry about permutations and compression and it just turns into a shitfest lol.
Once you have shadowmaps anyway it's probably more temping to just slap some PCF on top and not bother with writing another complex subsystem. Probably why no product ever shipped with this. But yeah, I just want games with faster shadow volumes :( The multiplayer gameplay looks very oldschool fast. Wish I had the machine to run it -
http://youtu.be/dWhsZAhfyqc but please, link the youtube vid in future. I hate giving kotaku traffic :(
http://www.youtube.com/watch?feature=player_embedded&v=Vxxnq5YAVHw Don't think is worth making a big news post for this, unless others think otherwise. Back in Feburary this year, I released a 3 level Wolf3D project. For anyone interested, here's some screenshots and a download link. Enjoy!
Screenshot 1 Screenshot 2 Screenshot 3 Screenshot 4 Download If I was going to defect from Quake it'd be Doom or Unreal that I would run to and not Wolf... Funny stuff though, never knew you could make wolf maps either.
A massive CPU behind it. Double that is half the noise time as it computes.
It's nothing feasible for the consumer market just yet, but in, say, 3 years, it'll be close to min spec. Assuming something better and faster isn't invented in the interim. Could be wrong about all this, but afaik the raytracing is done on the GPU. I also wonder how well it handles fully dynamic scenes, I'm sure there's some kind of acceleration structure in place to speed up the rendering.
Also, this is low framerate, low resolution and low samples per pixel. You'dd need much more than twice the computing power. I think more on the order of 16 for it starts to be feasible (twice resolution = *4, twice the framerate = *2, half as much noise = *2). A more likely road for future raytracing like solutions would probably be ray casting inside volumes like UE3 does, or do a dynamic unique virtual texturing that updates all the light data in texture space and uses rasterisation for rendering. That could be interesting as it also seems very well suited for a hybrid client/cloud gaming solution. I could imagine some indie game actually making an art style out of the grainy rendering by intentionally limiting it. Before hardware catches up, that is.
GPUs aren't designed to handle raytracing. The GPU can be used to supplement the CPU, but there aren't any 'acceleration' features in place.
There are some custom 'RPUs' that have been made, but obviously they are not common consumer hardware. http://en.wikipedia.org/wiki/Ray_Processing_Unit 21:59:34 | <onetruepurple> 320
21:59:51 | <onetruepurple> voreballs are 240, except for skill 3 where they're 340 and will outrun the player Scampie: Yeah but... there's already GPU raytracing stuff all over the place? Might be technical stuff I don't get.
I could imagine some indie game actually making an art style out of the grainy rendering by intentionally limiting it. Before hardware catches up, that is.
I sometimes wonder what if we had all the gpu power we have today but were stuck on 640*480 monitors. You could play some kid that lost his glasses, haha. Has to be the best resolution evar.
To me it will always be 'hi-res'. The moment when games were nolonger blocky as a mutherfucker.
|
|
| You must be logged in to post in this thread. |
|
| Website copyright © 2002-2024 John Fitzgibbons. All posts are copyright their respective authors. |