This is what we believe id is using in its next engine. Carmack stated something like "no two walls will look the same" so i believe that textures will use some kind of fancy tech to make them all look different.
Which wont matter a single fucking bit because they will still pump out a fucking boring sci fi first person shooter. Like Doom3/Quake4.
Yes, Carmack stated that procedural still wasn't cutting it and how Megatextures can be more effective and how it will likely be adapted on future engines for much more than just terrain (like on walls and stuff). But procedural textures != megatexture.
I'm serious. They look fake and bad, totally lack in style, and apart from one or two trivial examples fail at looking much like any kind of real surface. This is exactly the kind of cheap bullshit that can makes CG and games look so crap.
Carmack has still been referring to what they are doing for their new game as MegaTexture, which implies an evolution of that tech not a replacement for it. Unless he's come up with some miracle maths that makes procedural stuff look better than something a monkey shat out in 5 minutes in Photoshop, I'd be really surprised if he was using anything that's more strongly procedural than ETQW's terrain textures.
From what he's said, I got the impression that really not much has changed except that you can use MegaTexture on any surface.
I think procedural generation does have a place - if you take the example of kkreiger, that 64K demo file that ran a whole FPS level, you can see that its principles apply really well to games that have been created as small downloadables for delivery on something like XBox Live Arcade or Steam.
In its current state though, it does generally look terribly ugly and fake, so it's of relatively little use in prestige titles.
o'dium wrote:This is what we believe id is using in its next engine...
MegaTexture isn't 'proceedural' in the classic sense as it's not mathmatically generated (as is described in that article interview).
I believe what o'dium was talking about was not MegaTexture'ing but whatever id is doing in-house atm (i.e. - the tech *after* MegaTexture'ing).
Oh well we don't know if there's even going to be an 'after' MT (according the JC the next one is his last).
As Shallow said, the *next* licence is MT on everything (exactly how 'literal' that should be taken won't be answered until they tell us). What the *next* next project is, is anyones guess (but you can be sure they'll already be doing some R&D on it).
On the other hand, Carmack has said that his next engine may be his last. It has also been stated that the next IP they are working on will be a an evolution of the existing technology, not a complete change in the engine.
So whether or not the next game by Id will be considered his 'last' engine or just a progression of the current engine (meaning his actual last engine won't be part of this new IP) is also anyone's guess.
I think JC is very good at being ambiguous about his plans for the future, but regardless, I don't think he has any intention of retiring anytime soon.
some of those textures are pretty painfully obviously sourced from random noise. A nice step forward but still not replacing any dedicated artists anytime soon.
I've always thought some combination of the two would work best - get artist love to produce good looking materials and devise systems of procedural additions on top of them which are procedurally deployed based on geometry and designer application. How awesome would it be to just select "aluminum panel" as a material, throw it on some agglomeration of brushes, and have seams and stains and rivets automatically applied along edges and stuff? Or "metal floor" across a whole plane, and have the compiler divide it into plates, add metal trim and caution striping along the edges, and dirt/trash in the corners beneath walls?
I've been trying something like that in a q3 map, except by manually applying the decals myself all over the place, and it's a bitch. I might just go back to standard texturing for my own sanity.