Umm what's so incredible about using a particle emissions system?
A high density one would eat up fillrate like there's no tomorrow. Fillrate is a scarce ressource on RSX.
And please, please don't ever again say that the only recourse is fluid simulation (I'm guessing ur referring to fluid dynamics?). 2 points. The first is that fluid simulation is a completely useless term unless you specify the paramaters of what you are simulating, how finely, and what kind of simulation you are using (are you talking about a 100M molecule physics based sim?).
I wouldn't be using the the term fluid simulation (which is the correct term, by the way) if I wasn't refering to a complex fluid solver. Saying that it needs to be a complex is redudant, since you can tell a fluid animation solver from the good old textured large planes based stuff in a second. They're so obviously different in their implementations, their computational load and their visible look, that it goes without saying that they're not the same.
For those interested in the fluid simulation solvers, check
Nick Foster's work and
Ron Fedkiw's, for the academic and off-line side of stuff.
For the real-time solvers you can check :
http://graphics.cs.uiuc.edu/svn/kcrane/web/project_fluid.html
Keenan is the guy who worked the smoke box demo for the G80, he has the most impressive real time fluid simulation stuff going on at this time, that I'm aware of (and I keep myself up-to-date with these techs).
http://www.ss.iij4u.or.jp/~amada/fluid/
Old stuff, but interesting nonetheless.
Something worth of note is that I was dismissive about fire based on fluid solvers, saying that I didn't see one that was really impressive in real time, well after checking back Keenan stuff, I hav eto say that I either didn't pay attention the first time I saw it, or he did more work on it in the mean time, because it looks mighty fine now:
Be sure to check the video.
The 2nd is that fluid dynamics are certainly not at all necessary to implement decent looking fire. Volumetric effects like smoke are done in the same way as good fire is (good by our standards) in games like Crysis and many others.
What do you call volumetric effects? Because a simple particle emssion is a volumetric effect. Old PS2 games, such as Silent Hill 2, claimed to use volumetric fog, which was technically true, if we go by the definition that volumetric means, in a broad sense, that it has something to do with volumes.
More complex volumetric fog/particle systems, which uses raytracing/raycasting, are interesting only when they're covering most of the screen/camera space (fog/rain), and seem out of place for creating fireballs.
From the
old tech video I saw, Crysis just use a high density particle system, with the particles changing in time (from top to bottom), and it uses soft particles (an effect that blends the particles into the environment, which take care of the "card box passing throught the ground and object" effect).
And, in this case, after rewatching the effect again, I rest my case, it still doesn't look good. In other words, it suffers from the same issue Lair has, the fire textures do not look good, but where Crysis has to do deal with small entities, due to the high density of their particle system, and thus can't do much to the texture work on the flames without creating artifacts due to high frenquency textures, Lair, on the other hand, uses for their dragon's fireballs, more primitive, very large particles, therefore better/high frequency textures would look alright in this game.
It wouldn't scream "next-gen" effect, but it would sure look better than low resolution textures.
