Space Communist |
06-08-2011 03:09 AM |
Quote:
Originally Posted by LoBiSoMeM
(Post 294805)
Why not?
|
Well I suppose an argument could be made that as long as your gpu is up to the task then everything else can fall to your cpu. It could be that I am simply too used to thinking in terms of what has been true of gaming in the past.
But even so I think it would be a monumental task. Like in that video for instance, the game area appears to be quite large but I am willing to bet almost anything that a huge amount of what you see is not actually part of the game's world. Those mountains are almost certainly the edge of it, and I doubt they are navigable.
Take that smoke in the distant background in this video. Definitely amazing looking. However there is no way in hell that is some kind of volumetric fog. That is an animation that was rendered pretty much by hand and will play out the same way every single time. You can never actually get close to it or go into it, so it never really has to be simulated. It only takes a few resources from the GPU.
Once you provide that you might actually be able to interact with it though, suddenly you can't just use a graphical shortcut. you must now compute both its physical properties, and how those properties will translate into a visual. Herein lies the real problem: translating an effect that is simulated and variable into a visual is an order of magnitude more complex than simply using a stock visual that comes with a stock effect.
Anyway I will concede that it may be possible to get a simulation to look like this and still run on a top of the line machine. However it is pretty much going to involve having a budget that allows for things like "ok you are the shrub guy. Your entire job is to ensure that shrubbery looks good at any distance, from 3 inches to 10 km, with seamless transition in between, and that this shrubbery also interacts with all possible environmental objects and effects. Oh and make sure it is perfectly optimized as well."
|