[citation][nom]bogcotton[/nom]But so much of the data is exactly the same from frame to frame, there is just a slight shift of the image with a little extra info.If this was coded well I would expect only a 5% performance decrease frame for frame.It just seems software guys have lost their artistic passion for coding.[/citation]
3D is more than a shifted image, 3D tries to simulate stereo depth by simulating the imaging naturally done by your eyes.
With non stereo images, depth is done using shadows and highlights to give the look of depth, perspective can be added to further add depth.
in a stereo image; in real life each eye not only sees a shifted image, but you also get a slightly view angle of everything. When close up, you are able to easily see this if you close 1 eye but at a distance you may not notice this at all but within about 1000 feet, the brain can tell the difference and create full depth.
Stereo images require twice as much GPU work (while the images may look completely the same, everything is actually different (even flat objects, including the hud)
The only thing that 3D doesn't need twice of is memory, the same textures and other stuff loaded into memory is the same for both images
the GPU of the PS3 is based on the nvidia 7800 series and it is not much faster than the fastest 7800 available. the reason why we can have good looking games that we didn't see on the PC at the time is because the PS3 (and xbox 360) use a special trick to save GPU cycles for better graphics.
in the PS3 and most other current game consoles, when you look at a scene for example crysis 2
objects close to you are 3D objects with a full 3D mesh, distant scenery is flat (generally a giant sphere or cube surrounding the entire game world with a flat 2D HDR texture placed on it (so different times of day can be simulated from a single image.
clouds can either be flat moving objects if you are not required to go above them with out noclipping
or they can be just random particles surrounding a flat object with additional detail
areas where you interact with are made up of increasingly simplified meshes the further you are away from them.
texture resolution is also lowered for distant objects
dynamic lighting is only applied near by and pre rendered shading is used in the distance
this all allows much better graphics to be simulated, (at the cost of longer load times and and the need for faster system memory (which these consoles have very fast memory and CPU's )