Well, you can do remarkable things with speech synthesis these days. (Check out “Charles”—am I nuts, or does that sound like Roger Moore?) Even if you don’t want to do all the dialogue that way, it might not be impossible to set up a synth voice that sounds enough like a voice actor to passably say one word, and dub it in the appropriate spots in the dialogue. (Thank god you’re here, John Rambo! The orcs are breaking through the gates!)
Er, that last sentance was supposed to be in quotes, like a cheap demonstration. :smack:
Pretty much every one of those will be in The Elder Scrolls 4: Oblivion
It didn’t. It said it did, but it didn’t. You could blow up a small number of walls and things and it was usually pretty obvious.
As Ranchoth said, you wouldn’t voice-act every single conceivable name, you would synthesize it (either all of the dialog, or splice it into acted dialog). The real problem would be when players enter names like Xzypglskt or the like. Of course, the old text-based system where all the NPCs can flawlessly “say” names like Xzypglskt isn’t very realistic, either… Perhaps the ideal would be two separate dialog tracks, one where the NPCs make a strained effort to avoid saying your name? Then, you’d just need some sort of filter in the game to recognize “unpronounceable” names. If you’re really clever, you could make the filter vary from one NPC to another, such that for some names, a smart NPC could figure it out, but a dimmer one might not?
As for dedicated physics coprocessors, to a large extent, that’s what the graphics coprocessors are. Many of my colleagues rejoice at the advance of gaming technology, not because they themselves play, but because the graphics cards are excellent for many problems of real physics modelling. To the computer, it’s all number crunching, and it doesn’t actually care whether the numbers are crunched in service of pretty pictures or of boudary-value evolution partial differential equations.
The real problem with names is that a single string might map to two distinct names. Eg: Jesus might be pronouced “Gee-sus” or “Hay-soos”. The computer has no way to disambiguate between the two. What would be more feasible is for the gamer to first pronouce their name and then have the computer repeat the same phonemes using their voice.
I’m a game programmer, and I’ve been fascinated for years with the idea of fully destructible environments, but it (a) is technologically VERY difficult, (b) requires a large commitment in resources, both in terms of human development tmie and RAM and processing time, and © as others have pointed out, puts huge constraints on gameplay.
The game I worked on that got closest to it was a tank combat game in an arena, where the arena walls were impenetrable, and the matches were short, so if you went around and levelled every last thing in the arena, it didn’t ruin the game.