There's a host of interesting startup ideas in this field, most related to 3D graphics. We need higher levels of abstraction for creating textures, architecture/models, decoration etc. The tools we have today are awfully rudimentary, and there's research indicating that much better solutions are possible.
It is obvious that we could create a huge amount of tools to this purpose, and also that there is a huge market for software that makes game development easier. What's the sense in creating the 5000th 3D model of a human? Humans don't change that much. Not even fictional bipeds do. Have you ever heard a game developer or pundit complain about 'spiraling costs' for game development? It's because human developers have more and more work to do to make the games keep track with developments in hardware. It should be obvious to anybody that this development isn't sustainable; there's no way every room of a game designed by humans can have even as much detail as the room I'm sitting in right now. But most of this architecture is predictable, and could probably be generated by the right program.
Consider a lower bound of game development costs for this 'generation' of consoles at 3 million dollars. If even five percent of that cost could be eliminated by better development tools, that would mean savings (and potential for profit!) of 150,000 dollars. But the potential savings from automating menial modeling tasks are a lot higher.
Seriously. We need software to make stuff like this easier. There's a major can of worms in implications of sorting this out - a lower barrier to entry, more creative freedom, better-looking games and so on and so on...
See for instance http://www.vision.ee.ethz.ch/~pmueller/wiki/CityEngine/Docum... - a higher-level language for describing architecture. In the marketplace there are things like http://en.wikipedia.org/wiki/SpeedTree, which is already in use in a great number of games.
It is obvious that we could create a huge amount of tools to this purpose, and also that there is a huge market for software that makes game development easier. What's the sense in creating the 5000th 3D model of a human? Humans don't change that much. Not even fictional bipeds do. Have you ever heard a game developer or pundit complain about 'spiraling costs' for game development? It's because human developers have more and more work to do to make the games keep track with developments in hardware. It should be obvious to anybody that this development isn't sustainable; there's no way every room of a game designed by humans can have even as much detail as the room I'm sitting in right now. But most of this architecture is predictable, and could probably be generated by the right program.
Consider a lower bound of game development costs for this 'generation' of consoles at 3 million dollars. If even five percent of that cost could be eliminated by better development tools, that would mean savings (and potential for profit!) of 150,000 dollars. But the potential savings from automating menial modeling tasks are a lot higher.
Seriously. We need software to make stuff like this easier. There's a major can of worms in implications of sorting this out - a lower barrier to entry, more creative freedom, better-looking games and so on and so on...