“…real-time light/weight/sound physics that aren’t even integral parts of gameplay…”
Actually, you did say they weren’t integral =\
But nevertheless, in the instance of Thief: Deadly Shadows, is each individual shadow necessary for the gameplay? No. But, it blurs the line between “interactive objects” and “the rest of the environment.” Do you remember the days when in adventure games, you knew what items were interactive because they were drawn a little different? Same thing here. The developers were trying to create an environment that was alive and consistent. Why would some objects cast shadows, and not others? The more you blur the line, the more immersive the world becomes. As I mentioned to Gehn in a separate thread, “consistency” is what you need to strive for in a game.
And when you have a ton of stuff to render, it’s actually easier to cast dynamic shadows, rather than draw everything in yourself. The act of making everything have shadows isn’t what’s causing costs to go up (although the initial R&D does), it’s the fact that now that games are expected to have a ton of content. If you were to go in and try to draw in the shadows all by hand, that’d be insane. So, it costs extra money during R&D to develop vertex shaders (unless you license the engine), but it allows developers to devote time and efforts elsewhere during the actual building of the game.
And did you ever think that bump mapping was what actually allowed them to build the game the way they did? It wasn’t that there was an incredibly low poly count… they just had to divide the polys up between a whole lot of objects and models. Bump mapping allowed them to accomplish this without sacrificing too much quality. Normal mapping allows this to an even greater degree.
You do realize that, for example, the wheels on the Warthog in the original Halo were actually octagonal, or something crazy like that? Bump mapping is what allowed them to look like they were actual wheels.
These are far from graphical gimmicks, as they build on the believability/consistency of the game world. And unless you do something as a developer to at least stay above the minimum acceptable technology line, you become irrelevant. People take one look at Doom 3 or Half-Life 2 and say, “Man, your game looks like crap compared to these.” And that’s because it does.
As far as Halo goes, the developers made the game with giving the player freedom in mind. As it goes, the player could be standing in any place within the level at any given time. To give it the most realistic sound possible, they created the system you see now. Is it the most efficient? Probably not. But is it the most effective? Most likely. With an interactive medium, there will always be wasted efforts because the player will never do exactly what you want 100% of the time. Such is the nature of the beast. Actually, this is one of the very reasons why content generation for games is becoming more and more of a monstrous beast. People expect RPGs to play like Morrowind/KOTOR/Fable now. Technological accomplishments of these games aside, the simple act of generating that much content, simply because these games don’t play on a linear path like FF does (or to a lesser extent) makes a lot of what goes into these games superfluous to a lot of gamers, yet it’s gamers themselves that clamor for more games exactly like that.
There are plenty of games out there that feature new-fangled technology or whatever that also have great gameplay. But, the companies that are going to be able to afford to do all this are going to become fewer and fewer until something drastic changes in the way games are made/published.