I'm excited about the potential in procedural generation. And in user-made content. But I haven't seen anything that does a really fantastic job yet. Aesthetic isn't easy to program.
For realism, it'll be useful. But I doubt it'll ever be truely useful to people like Blizzard. Or for games like Okami.
For that, you need really good tools. Which we don't have right now. The best proof of this is feature films. 3D animation is always of a far less quality, animation and design-wise, as 2D animation. This is because the artists doing the work have to work against very hard-to-use and hard-to-learn tools. So it is a different segment of animator that uses it, a smaller sub-section, and one with far less animation talent, and far more technical talent.
Tools can be made to do a lot of work, one of the biggest jobs I hope to see all but disappear is the texture-artist. A proper high-poly modelling software, with good tools, should be able to alter textures, procedural and photographic on the fly... I've seen apps that make photos tile perfectly and seemlessly, Maya should do this automaticly, and be able to 'fix' bad textures, randomize them and even use them as a guide to build never-repeating mega-textures based on them.
It should have a streamlined materials section that has millions of pre-set types, and if they don't match right, they should be easily fine-tuned. The system these programs use is counter-intuitive and designed by and for programmers.
If they JUST did what I said here.. You'd see at least a 20% faster development time for models... perhaps even better than that.
If they fixed all of the terrible interface issues and AWEFUL bugs... (in both maya and 3dstudiomax) Then I'd expect nothing less than a 70 or 80% time reduction.
Large-poly models should reduce to small-poly in a single button hit. Retain all the details in the high-poly model, turning all the lost geometry into alterations to the texture.
There should be some really powerful level-design tools. Specific to level-design. Think, WC3 editor, but for anything. With the ability to read and output into any format as defined by the designers. So you don't work around the format, it works around your game.
Even a lot of programming could be reduced to nothing. While programmers pride themselves on their ability to fine-tune things. There should be powerful tools, not unlike blitz, that really reduce dev-time. Perhaps not even language based.. Could be very wysiwyg even.
One of the areas that procedural models would be most useful is in getting 'close'... Say, you give it some textures, the basic idea of what you are making ... example 'mushroom' and it'll output a bunch of mushrooms. You pick whatever ones you like, and then edit them more to get them just right... Something like this could DRASTICLY speed up production.
I mean... we are talking about reduceing a dev-team from more than 200 down to 15-20. And them being able to get things done faster.
Then you can spend all your time making the best game you can... Not making content.
Something like "Band in a box" could be used to make music for a game. With a tracker-enabled version. It could make professional quality music for games... You could even then have music custom-made for the situation the player is in... as he gets in it. At least eventually.
There could be more research into off the shelf AI, really super-advanced and fine-tuned AIs that you can plug into any game. Then you can focus on the stuff that isn't the same in each game.. or focus on altering the minor things that are different.
Lets see.. what else is there...
For realism, it'll be useful. But I doubt it'll ever be truely useful to people like Blizzard. Or for games like Okami.
For that, you need really good tools. Which we don't have right now. The best proof of this is feature films. 3D animation is always of a far less quality, animation and design-wise, as 2D animation. This is because the artists doing the work have to work against very hard-to-use and hard-to-learn tools. So it is a different segment of animator that uses it, a smaller sub-section, and one with far less animation talent, and far more technical talent.
Tools can be made to do a lot of work, one of the biggest jobs I hope to see all but disappear is the texture-artist. A proper high-poly modelling software, with good tools, should be able to alter textures, procedural and photographic on the fly... I've seen apps that make photos tile perfectly and seemlessly, Maya should do this automaticly, and be able to 'fix' bad textures, randomize them and even use them as a guide to build never-repeating mega-textures based on them.
It should have a streamlined materials section that has millions of pre-set types, and if they don't match right, they should be easily fine-tuned. The system these programs use is counter-intuitive and designed by and for programmers.
If they JUST did what I said here.. You'd see at least a 20% faster development time for models... perhaps even better than that.
If they fixed all of the terrible interface issues and AWEFUL bugs... (in both maya and 3dstudiomax) Then I'd expect nothing less than a 70 or 80% time reduction.
Large-poly models should reduce to small-poly in a single button hit. Retain all the details in the high-poly model, turning all the lost geometry into alterations to the texture.
There should be some really powerful level-design tools. Specific to level-design. Think, WC3 editor, but for anything. With the ability to read and output into any format as defined by the designers. So you don't work around the format, it works around your game.
Even a lot of programming could be reduced to nothing. While programmers pride themselves on their ability to fine-tune things. There should be powerful tools, not unlike blitz, that really reduce dev-time. Perhaps not even language based.. Could be very wysiwyg even.
One of the areas that procedural models would be most useful is in getting 'close'... Say, you give it some textures, the basic idea of what you are making ... example 'mushroom' and it'll output a bunch of mushrooms. You pick whatever ones you like, and then edit them more to get them just right... Something like this could DRASTICLY speed up production.
I mean... we are talking about reduceing a dev-team from more than 200 down to 15-20. And them being able to get things done faster.
Then you can spend all your time making the best game you can... Not making content.
Something like "Band in a box" could be used to make music for a game. With a tracker-enabled version. It could make professional quality music for games... You could even then have music custom-made for the situation the player is in... as he gets in it. At least eventually.
There could be more research into off the shelf AI, really super-advanced and fine-tuned AIs that you can plug into any game. Then you can focus on the stuff that isn't the same in each game.. or focus on altering the minor things that are different.
Lets see.. what else is there...