in a thread a month ago about doom3 unified lighting came up. (ie each surface is treated with the same methods)
from my understanding this means
eg every object/polygon has a stencil shadow volume (from the screenshots this doesnt look to be happening ie not all objects are casting stencil shadows)
(of course volumes are used with projected textures as well as stencil shadows)
also for those who mightnot of seen it http://www.gamespy.com/articles/august02/quakecon2002/carmack/
seems like hes already thinking about the followup to doom3. personally from what i think it is doom3 lighting is a passing fad (which im skipping)
I just think it means that every light illuminates every surface it sees with the correct bump mapped lighting equation and the results are accumulated in the framebuffer. This is perfectly unified. Yes objects should cast shadows but I don’t think there should be constraints on HOW they cast shadows or even if they cast them as a matter of optimization. People seem to read what they want into the ‘unified’ description.
It’s interesting that Carmack talks about the flare effect that’s used in the Wreckless game, and subsequently demonstrated in Nutty’s little demo. I really believe this effect adds tremendously to images - I’ve got it as a togglable in my ‘engine’, and it looks great!
This could turn into a long thread…
In the GeForce3 launch video, Carmack said the lighting is now unified. What is not unified according to Carmack, is doing per-vertex lighting for the characters, lightmaps for the world and “shadows that are done a couple of different ways for different things. Everything is now done in exactly the same way”. Those were basically his words.
Well, I have no doubts .
OK dorbie youre saying only the lighting is unified BUT shadows aint.
sorry by lighting i meant ‘illumination’ ie combination of lighting + shadows (lack of light).
is illumination the right word?
OK shadows are caused by light not falling on a surface. thus if the lighting model is only illuminating the surfaces ‘in view’ then shadows come automatically.
are there shadows (as in actually created like all the stencil buffer shadow demo’s on the net do eg nvidia ones) in doom3?
or are they done with the surface recieving no light (btw this is what that small man in room demo i done a month ago used)
none of this explains though why the doom3 screenshots DONT look like a unified lighting model eg creature shadows are darker than shadows from objects (thats if they cast shadows in the first place)
excuse my inabilty to explain myself (i have problems transferring my thoughts into writen words)
I’m not really clear on this. Is everyhting gone be done the same way -> stencil shadows and GL lighting, or is it a combination of technics.
or maybe by unified he means you can have lightmap + cheap shadows or lightmap + stencil shadows or … and that is handled by one part of his engine.
And he says he was aiming for Geforce 256 with Doom3, so I think a combination is used. I wonder what it would look like playing on a lower end card. All those sweet effects will have to be turned off.
Well like I said people read what they like into ‘unified’. I’m kinda surprised someone is saying that lighting is per vertex on the characters. They are bump mapped so at least the dot products must be fragment lighting, maybe it’s some attenuation terms that are per vertex.
I have lower expectations of the ‘unified’ phrase I guess. I don’t see the use of multiple shadow algorithms as being non unified. I see unified meaning everything works consistently and accurately in a single system without compromises. I always thought it used soft & hard shadows.
What I thought was crazy in the long doom3 thread was people saying “No it can’t do it that way because Carmack says it’s unified.” as if one word open to interpretation determines everything.
Now interesting possible discussion. Vertex lighting on characters, did Carmack say this? What do you think is computed per vertex on the characters? Clearly normals are done per texel. So what & why per vertex or is this a misquote?
[This message has been edited by dorbie (edited 08-25-2002).]
you read it wrong dorbie:
What is not unified according to Carmack, is doing per-vertex lighting for the characters, lightmaps for the world and “shadows that are done a couple of different ways for different things. Everything is now done in exactly the same way”. Those were basically his words.
the vertexlighting and such is done normally, and that is NOT unified lighting he ment.
OK fair enough, so he was just using that as an example of what would be non unified in earlier engines. Not saying that’s how it’s done in Doom3.
Yes, I was just giving the example that Carmack gave on what the lighting/shadowing was like inn the old engines ( lots of different ways for different things ). Now everything is bumpmap lit with stencil shadows. My point was that all effects that the DOOM engine can do, can be applied to all surfaces ( not just a subset of surfaces, like the characters ).
Actually, you know I agree with you on the ‘unfied’ term ( from a previous thread ).
Just saw a new interview with Tim Willits,
Here’s a snip,
Because of the unified lighting engine all objects, regardless of where they are made (brushes and models), look identical in the game.