I'm going to qualify this thread as my opion, first and foremost. 
It's also in response the the great amount of RSX vs. Xenos threads I've been seeing lately.
I've observed few things over the past console generation with regards to games tailored to architectually different machines. The thing that really sparked this thread is many of PS2's amazingly cinematic, heavlily CPU-reliant games that have been produced over the past 2 years, namely God of War, SotC, MGS3, DMC3, etc. The ability to use a graphically weak machine like the PS2 with it's lack of any shader capability and tiny pool of RAM and make games that look better moving than games utilizing the most cutting edge PC technology says alot for how graphics are more of an artistically dependent matter and less a technically limited matter on the GPU end. If anything, a CPU-centric machine churning out fluid animations, natural transitions between said animations, and good collision detection running at a high framerate, coupled with classy artistic style, makes for games that more closely resemble prerendered movie visuals than games with stiff animation and bad collision detection coming from a more GPU-centric configuration that spits out a pretty picture in a dead environment.
What I'm getting at is that I'd rather see less sophisticated in-game assets move and interact in a more sophisticated manner with a vibrant, living world than technically sophisticated models, shaders, textures, landscapes. In screenshots, most people eyes pick up on good art before technical sophistication of models, anyway. And in motion, good art, animation, etc. are really the things that matter on a visual level. The next logical progression from here to add meaningful interactive elements on a visual level would be to establish good physics-based animation and simulation and include smarter AI, all being heavily CPU-reliant activities. Thus, I think the public's expectation of next generation gameplay and visuals (without them even realizing it) is a more naturally flowing game, not a game with just more complex models and environments. Programming on the CPU end should be the focus of progress in this next generation, with GPU programming being secondary. Otherwise, the uncanny valley will be here to stay.
Couple this mode of thinking with better game design approaches that suitably balance interactivity with technical limitations (see David Jaffe and the God of War team) and we will be arriving at Next Generation gameplay sooner than if we made the focus on writing more sophisticated shader programs. What is your (everyone else's) opinion on the matter?
It's also in response the the great amount of RSX vs. Xenos threads I've been seeing lately.I've observed few things over the past console generation with regards to games tailored to architectually different machines. The thing that really sparked this thread is many of PS2's amazingly cinematic, heavlily CPU-reliant games that have been produced over the past 2 years, namely God of War, SotC, MGS3, DMC3, etc. The ability to use a graphically weak machine like the PS2 with it's lack of any shader capability and tiny pool of RAM and make games that look better moving than games utilizing the most cutting edge PC technology says alot for how graphics are more of an artistically dependent matter and less a technically limited matter on the GPU end. If anything, a CPU-centric machine churning out fluid animations, natural transitions between said animations, and good collision detection running at a high framerate, coupled with classy artistic style, makes for games that more closely resemble prerendered movie visuals than games with stiff animation and bad collision detection coming from a more GPU-centric configuration that spits out a pretty picture in a dead environment.
What I'm getting at is that I'd rather see less sophisticated in-game assets move and interact in a more sophisticated manner with a vibrant, living world than technically sophisticated models, shaders, textures, landscapes. In screenshots, most people eyes pick up on good art before technical sophistication of models, anyway. And in motion, good art, animation, etc. are really the things that matter on a visual level. The next logical progression from here to add meaningful interactive elements on a visual level would be to establish good physics-based animation and simulation and include smarter AI, all being heavily CPU-reliant activities. Thus, I think the public's expectation of next generation gameplay and visuals (without them even realizing it) is a more naturally flowing game, not a game with just more complex models and environments. Programming on the CPU end should be the focus of progress in this next generation, with GPU programming being secondary. Otherwise, the uncanny valley will be here to stay.
Couple this mode of thinking with better game design approaches that suitably balance interactivity with technical limitations (see David Jaffe and the God of War team) and we will be arriving at Next Generation gameplay sooner than if we made the focus on writing more sophisticated shader programs. What is your (everyone else's) opinion on the matter?
