Jump to content
Search In
  • More options...
Find results that contain...
Find results in...
couvier

The 3D engines are getting more real

Recommended Posts

All we need now is radiosity lighting with real time moving lights/objects, which Doom3 does not have. So Doom3's lighting is still not "true" lighting yet, but it's a step closer. It's interesting to realize that quake3 has 1 lighting advantage that doom3 does not have, BSP compilers calculated radiosity for the final product map, but of course doom3 has so many other lighting advantages that outweigh quake3's. It's not John Carmack's fault that we dont have radiosity in Doom3, it's simply that CPUs/GPUs are not powerful enough to compute radiosity in real time with moving objects yet. Were basically waiting for nVidia and ATI to get graphics cards out that can handle it.

This link below shows you guys what the future of Carmack's engines will do, that Doom3 does not do. Doom3's engine is the pictures that say "No radiosity" on it, future engines will be the ones with radiosity included.

http://www.students.tut.fi/~warp/pics/Radiosity_test2/

The last example is my favorite one, the flashlight is shining at the corner of the ceiling, no radiosity shows complete blackness behind the flashlight, with radiosity included, light reflects off of surfaces (the ceiling corner) and bounces around (like real life light physics). Light reflects off of light colored surfaces (white), and gets absorbed by darker colored surfaces (black). So how much light reflects depends on the diffuse surface.

I’d also like to note that radiosity lighting generally makes environments “brighter” and more “real life”, probably one of the reasons they chose to make a single player doom game with the doom3 engine and not multiplayer, is that the engine suits it best. It makes things more darker, gloomier, doomish, etc.. Since things will be darker, a multiplayer deathmatch engine will probably have to wait till radiosity lighting will be included with Carmack’s engines (so brighter environments for concentration on fragging, like quake3). Maybe we will se it for the quake4 engine.


Other radiosity comparison links:

http://www.students.tut.fi/~warp/pics/Radiosity_test/

http://o.ffrench.free.fr/graph/radiosity/

http://www.google.com/search?q=Radiosity (just look around, you’ll find tons of stuff on it)

Share this post


Link to post

Radiosity would be nice, but the calculations for it require a ridiculous amount of power.

It's easy to render it for a still image in a 3D scene, but do it in real time, with dynamic lighting and you run into a bit of a number crunching problem.

From the comparisson photos it does look more realistic than hard edged shadows, but at the moment even the Ti4600 doesn't have enough power to render a scene without the game becoming a slideshow.

I'm just happy that Carmack is putting in real time shadows, the difference in realism from lightmaps is just stunning.

It's like the difference in lighting from Wolf3D (no brightness adjustment) --> Doom (single sector lighting values) --> Quake(s) (lightmaps) --> Doom 3 (stencil shadows)

Share this post


Link to post

2 or 3 years? I don't think so. While there's a big chance the big companies might be able to release that kind of tech, I don't think developers would like to sacrifice detail for such feature. I'd rather see even more 2x, 3x more complex scenarios than just a nifty addition to the lighting.

Share this post


Link to post
Zaldron said:

I don't think developers would like to sacrifice detail for such feature.

Why do you think they would? It's just an engine feature, has nothing to do with their mapping layouts and whatnot.

Share this post


Link to post

Well, by the time we have the technology to make Doom III level of detail work with radiosity in real-time, I'm sure developers will be doing far more detailed areas, hence it's still slow as fuck.

We need some kind of technology breakthrough first...

Share this post


Link to post
HAMMER-STROKE said:

we need a Quantomcomputer

No they're not anything special for gaming.

Share this post


Link to post

The deal with quantum computers is they're pretty elite at finding complex patterns in multidimensional arrays and such, but they're not extremely good for dividing polygon angles at ten million hertz :P

Share this post


Link to post
Zaldron said:

2 or 3 years? I don't think so. While there's a big chance the big companies might be able to release that kind of tech, I don't think developers would like to sacrifice detail for such feature. I'd rather see even more 2x, 3x more complex scenarios than just a nifty addition to the lighting.


If you remember the transistor law, in 5 years we could be at 62 ghz if things continue the way they are. That feature is not impossible

Share this post


Link to post

I read in a pc magazine .. that we would be up to the 100ghz in 5 years, because of transistors created with some sort of nanotech.. or something like that

Share this post


Link to post
HAMMER-STROKE said:

I read in a pc magazine .. that we would be up to the 100ghz in 5 years, because of transistors created with some sort of nanotech.. or something like that


Your talking about nanotechnology, but in 5 years, nanotechnology won't be fully developed and available for home pc use. I don't think we will rise in ghz but more like better arcitecture. The Intel Italium and AMD Sledgehammer both run very low mhz but have better arcitecture then the p4 and Athlon XP and run a lot faster

Share this post


Link to post

Well, you could always use precomputed radiosity from a program like lightscape. It allows you to edit the radiosity solution and save it as textures. http://www.autodesk.com/us/lightscape/examples/html/index.htm
Too bad that no game makers ever used it (as far as i know). The disadvantage is that it wouldn't be dynamic and consume lots of memory (every surface needs a unique texture), but it would look more photoreal than doom3 (still not half as cool though).

Share this post


Link to post

Too bad that no game makers ever used it (as far as i know). The disadvantage is that it wouldn't be dynamic and consume lots of memory (every surface needs a unique texture), but it would look more photoreal than doom3 (still not half as cool though).


Uuh, that's pretty much what lightmaps are.

Share this post


Link to post
Fredrik said:

The deal with quantum computers is they're pretty elite at finding complex patterns in multidimensional arrays and such, but they're not extremely good for dividing polygon angles at ten million hertz :P

The deal with quantum computers is... that there has been no significant progress on it. I think the latest thing they made was something that could calculate 2 qubits... and even then I don't think it worked lol.

Also

The 3D engines are getting more real

lol duh

Share this post


Link to post
Psyonisis said:

The deal with quantum computers is... that there has been no significant progress on it. I think the latest thing they made was something that could calculate 2 qubits... and even then I don't think it worked lol.

No I've heard of at least two achievements they've done. For one, they added two single-digit integers and it worked. :)

Share this post


Link to post
DaJuice said:

Lightmaps are generally made with lightscape??


No, but lightmaps are the same basic concept, just created by the map compiler...

If someone were going to use lightscape to create lightmaps, they would have to import the map into lightscape somehow. That would just be a pain and more work than it is worth.

Share this post


Link to post

First of all, Carmack toyed with radiosity in Quake2, not Quake3. He decided the hard-edged "simpler" shadows of Quake1 were better because they were cheaper to calculate and the players didn't seem to care about the radiosity in Quake2. So he ditched it with Quake3 and stuck with a simple lighting algorithm.

As for radiosity in realtime, it's going to be a while before it ever comes to a reality if we follow the path we are currently on.

It's hard to explain but radiosity goes alot slower than most people think.

You could do a quick-hack radiosity where light can only rebound once off a surface onto another surface once but that would still be exponentially slower for each triangle you add into the scene.

Radiosity is a ways off if we keep doing things the way we are now (eg: Doom3).

Share this post


Link to post
Guest
This topic is now closed to further replies.
×