a very interesting question i have had recently. Actully, this springs from my knowllege of 3d modeling for movies. which do you think is better? texture mapping on a flat surface or material filling on a modeled surface. What i mean is, we all know that in the movies, no mapping is used as no surface is really flat... with the exception of walls of coarse. WE might normally model the surface with all its pits and surfaces and then paint it. with game art, we just apply a simple texture and this does the job. we might add bump mapping or something to make it more real, but i ask, in an industry where we have radeon or g-force, why is it not possible to make a world with real work modeling but instead of tilable textures, we coat it in painted materials.. in a sense, these materials are still textures but they would not have to simulate bumps or imperfections like we expect our textures to do.. so, with the ability to run millions of pollies per second, why not? i ask this because im really sick of seeing simple bring walls that light hits the shadow of dirt that looks the same from all angles. i want to see ritch grass where each strand is modeled and each one behaves on its own. i want to see trees that are real world objects and not branch textures mounted on a tube. maybe a brick wall that is actully made of brick. on top of this, character modeling that makes clothing out of real 3d cloth. this relates to my atomic code filling but who knows. im probably stupid and know nothing... im just not impressed with the claims at realistic looking 3d graphics that still look like games.. I am not text, I am not organized pixels, I am not killed by turning off your monitor, I am not isolated by turning off your computer. I just am.

Conshape Electronic Arts Millennium ED Redwyrm Online

I think the new shaders in DirectX 8 are the first step to letting us do what you call material filling. Although we are a long way from being able to render many blades of grass in realtime.

Texture mapping won''t go away anytime soon though, if ever.

--
Todd
http://www.3dcgi.com/

i think it will. basically, we are playing with more sophisticated doom when we do our 3d graphics. eventually, i dont think that 3d worlds will be decernable from real photographs.

I am not text, I am not organized pixels, I am not killed by turning off your monitor, I am not isolated by turning off your computer. I just am.

Conshape Electronic Arts Millennium ED
Redwyrm Online

not necessarity.
The object of visual presentation is to stylise the visuals.
Quake III Arena is not realistic as such, its got a style, which can even by seen in the brief DooM 3 footage.

Absolute realism is some games would be very droll.
One of the wonderful things about CG is you aren''t restricted to real world perseptions or rules.

You can do anything in CG and style is one of the factors you can incorpate.

Flight Sims would benefit from absolute realsim since that is what they arew simulating, and the same could be said for MechWarrior, since the would look very cool with realworld conditions

But sometimes it''s not best to go out and photograph a wall to make your game "realistic" sometimes it''s best to have charactericts with the art.

Diablo and Starcraft would never work as well with ultra realism. (IMO)

oh by the way, I think there are ways to do unique walls reatime, but memory becomes an issue then due to the various textures required.

Radeon and GeForce are nowwhere NEAR powerful or versatile enough to handling the detail you''re talking about. For rendering detailed scenes, the problem at the moment is overdraw. The Doom 3 engine for example with have a possible overdraw of 40 million polys. For each frame your would "draw" 40 million unseen triangles.

Materails in games? Well it would be nice but memory again, and no 128mb is not enough! (you''ll want texture 2048 x 2048 for the realism at least, compared to 256x256 64MB ain''t going to cut it.) Also colour depth we need at least 64bit colour, perhaps favourable more, since the more passes you render the more it affects the colour. (remember "banding" with 16bit colour? well thanks to realtime light, fog, mist and particles 32bit is will start to struggle.)

Imagine looking through your visor, at a stain glass window, behind which is a foggy gravyard with a spell weilding skeleton is hacking at a guy with a torch? because of all the surfaces light and transparancy the colour is affected because of the 8+ layers of fx in one area, the precision is not there to get the exact right colour req.

"i want to see trees that are real world objects and not branch textures mounted on a tube. maybe a brick wall that is actully made of brick. "

Btw, if you were working on a game, woul you like to be "artist" or "bricklayer"?

seriously though, you do not need to build a brick wall via each brick, there always a faster way. (and time is money, if you build a scene brick by brick so to speak dev time would exceed 5-10 years! who''s going to fund that? is it going to be more fun because of the fact each wall is build from separate bricks? nope.)

Technically, if you reeeaaaallly wanted to, you could now, but it''s not necesary.

Gotta hand it to you on trees tho, you''re absolutly right that current techique is, pardon my french, shit.

a polygon bark with 4 transparent square textures in form of an "X" is one way to do trees realtime atm. Though with GeForce 2 and 3 cards and Kyro II chipsets you can render tree with branches and leaves. So, slowly but surely, se''ll get there...

In a nutshell:-
CG is cool, and does not have to be relaistic (it''s not what all games are striving for) - we are a long way off doing realistic stuff realtime. (We can do Luxo Junior is realtime tho - Pixars first animation)

I''ll stop typing now.

www.stephen-hawes.co.uk