DuckThor Evil
Legend
- Status
Is this game based on this method?
http://portal.acm.org/citation.cfm?...l=portal&dl=ACM&CFID=15151515&CFTOKEN=6184618
http://portal.acm.org/citation.cfm?...l=portal&dl=ACM&CFID=15151515&CFTOKEN=6184618
I don't see any global illum or occlusion stuff in there. It's a nice soft ambient lighting, probably some ambient cube map based stuff, but nothing more. Half-life 2 started it, Halo3 and other games are doing it, and so on.
LBP's rendering technology doesn't seem to be too impressive to me, it's how they put it to use that makes it look nice.
LBP's rendering technology doesn't seem to be too impressive to me, it's how they put it to use that makes it look nice.
Yeh I just saw that pdf and it's the Global Illumination approximation they're using. Very clever stuff. It's the reason why it all looks solid and real. I hope more games can utilise this technique.
Please ,take some time to read the paper....
And show the video of the SIGGRAPH 2006 presentation of Alex Evans on the link given by one:
http://forum.beyond3d.com/showpost.php?p=947448&postcount=103
Didn't see that video first time through. There's a little demo toward the end that is definitely 'LBP' in early form. He throws around a lot of light-emitting-stars with a big-headed avatar 
He talks about how his algorithm allows for many many lights..something which they mentioned at the GDC demonstration indirectly (i.e. that players could create dynamic light-emitting objects arbitrarily and put them wherever they want). Another constraint he talks about is limited depth complexity..the slide even mentions '2.5D'
He talks about how his algorithm allows for many many lights..something which they mentioned at the GDC demonstration indirectly (i.e. that players could create dynamic light-emitting objects arbitrarily and put them wherever they want). Another constraint he talks about is limited depth complexity..the slide even mentions '2.5D' 
Arbitary lightsources casting soft shadows with illumination feedback isn't impressive?
The .mov mentions using the third method in a shipping title. It starts at about 30 minutes in. 36 minutes shows the bighead character with multiple lightsources, and the GI-look is very apparent.
What technology?
Yet another correction after watching the video - it sounds like LBP is using a technology that's not in the slides, and it "only" does lighting with an arbitrary number of lights. No global illumination at all, no occlusion either, not even in fake form, as it sounded.
I also recall nAo mentioning that he has recognized a certain type of shadow maps, so that also suggests "traditional" shadows, instead of fake AO or similar techniques.
I'm gonna have to re-watch the LBP video presentations though...
I also recall nAo mentioning that he has recognized a certain type of shadow maps, so that also suggests "traditional" shadows, instead of fake AO or similar techniques.
I'm gonna have to re-watch the LBP video presentations though...
Last edited by a moderator:
This is the best quality HD video of the demonstration I've seen so far:
http://www.gametrailers.com/player.php?id=17723&type=mov&pl=game
Looks incredible.
http://www.gametrailers.com/player.php?id=17723&type=mov&pl=game
Looks incredible.
The lighting in LBP is distinctly more realistic than any other game I've seen, so it must be doing something very different from other games, and probably very close to the GI and AO approximations in the Evans papers. I can see that the AO is not perfect sometimes (e.g. colour bleeding through solid objects), but approximations cannot be expected to be perfect, and their flaws should not be used to suggest that they're not being done at all.
Phat
Laa-Yosh, how do you come to the conclusion that LBP isn't using View Aligned Irradiance Slices? The demo shown exhibits many characteristics from the LBP trailer, and the algorithm only works for "small world" (low depth complexity) constraints (2.5D) which is exactly what LBP is.
Sure, it's not a traditional GI algorithm, but the fake look is incredible. Anyway, if you can render a very large number of lights scalability, you can approximate GI but making every surface one or more light sources.
Sure, it's not a traditional GI algorithm, but the fake look is incredible. Anyway, if you can render a very large number of lights scalability, you can approximate GI but making every surface one or more light sources.
It does use the irradiance slices; what it does not seem to use is the signed distance fields hack for AO-like soft shadow effects. People seem to get these confused already, because the downloadable slides only mention the SDF stuff, but the whole Siggraph presentation demonstrated 3 different techniques and some assume that a different one - or all three - are included in LBP.
To me it seems that only irradiance slices are used, to enable lots of lights without a drastic and variable - thus unpredictable - computational overhead. That's how there are many little lights in there at the same time.
To me it seems that only irradiance slices are used, to enable lots of lights without a drastic and variable - thus unpredictable - computational overhead. That's how there are many little lights in there at the same time.
Tentative release date is.......?
- Status
