"low-quality rendering"

strange idea, but I doubt you'd be able to get away with it in more than one game or so. This would be great when hi-rez VR headsets are found in every home, or we are all cyberized (think Ghost in the Shell) playing games in our "mind's eye" through implants.

And the reason I don't think you'd get away with it is because we are used to looking at each and every corner of our computer monitor, and even playing a FPS, I dont move the mouse yet i move my eye to the portion of the screen where the action is occuring first(before centering on it with the mouse). And can you imagine making more than one game this way, the "hardcore" review sites would tear you a new one for poor quality cuz their 16xFSAA quad-SLI machines are not giving them crisp pictures ;)

However if you make a horror FPS, I can see this method rendered in two passes making an oval mask with the lower rez render on top of the center rendered hi-rez geometry, then heavily filtering the low-rez render and "viginetting" darkening the edges inward with the pixel-shader, to make the effect seem more intentional.

But then you would waste so much PS power blurring and post processing, that the geometry in the center would need to be supremely hi-rezolution with millions of polys and hudreds of shaders for it to be worth using. Not to mention the lag incurred that changing render targets, then passing through your scene's "scene graph" visibility tests twice(I suppose you could cache results from the first run, but what an added pain that is), and many other redundant actions.

Oh and the biggie, geometry is now the (relatively)cheap in terms of detail, what is expensive is the ammount of texels sampled and the pixel processing performed on those samples to output to visible pixels.

The only thing I can think that would work, is if you are using mip mapped textures you could lower the level bias in view-space as you radiate from the center. That might be a bit cheaper for the texture sampler to use a lower mip level, though I dont see how, I have never tried. And it might actually be an interesting effect in the end, but again just a novelty feature to be used in one game and its sequels.

hopefully that rant made some sense :)
-Dave

It's not a terrible idea, but if you think it through, I believe you'll see that there are some inherent problems with it. I've thought about the idea of slightly blurring just the very edges of the screen to provide some illusion of focus, but I've never considered that as a way of saving rendering time. And people have done considerable work on providing 'focus' in one way or another, which is sort of a variation on what you're talking about, although again it's not done with any expectation of saving cycles, and to the best of my knowledge is not terribly useful in a game or any virtual world as it stands now, because people really don't want to bother to, for example, constantly move their mouse to be hovering over the object in focus.

In any kind of a game, even an FPS, people will, at some point or another, look at the edges and corners of the viewable area for some reason, rather than always moving the view to put their primary focus in the center. Look at your screen from a comfortable distance - say a foot and a half. Unless you have a really large screen (and you can't assume everyone does), then it is only filling a fairly small portion of your actual field of vision out to the periphery. If you render the edges of the screen at a lower detail (or blur them or whatever), all you will really have succeeded in doing is to reduce the useful size of your screen.

As long as the monitor screen is flat (relatively) and of limited size, then there will always be a bunch of inherent limitations as to how real it can look, and how much you can do to overcome those limitations.

Personally, i think it's a terrible idea. You haven't given details so i don't know what you have in mind specifically, but if you're thinking of using lower textures resolution or more simple shaders in the corners of the screen, you'll have to face the obvious problem of popping (ie. you're rotating the view and suddenly your object with low res textures and simple shaders enters the "high quality" zone). To get ride of popping you'd have to use tricks and linearly interpolate between the low and high res textures, or interpolate between the low and high quality shaders; which means the high res textures or high quality shaders still have to be executed, even in the corners of the screen. Net result is both a loss of performance and quality.

Y.

As a gamer, I should say that I wouldnt like the idea. I like looking around with my eyes. Its much easier than to look around with the mouse. And I like to stand in the corner, and camp without moving the mouse... you know, 90° view frustum. That wouldnt be possible with the "you can only see the center" games.

But in an UT-kind game, it yould be good. Extremely nice-looking effects in the middle, and low-quality stuff on the edges.

Interesting idea. I think the theory would apply better to streaming systems. When an object first comes into the FOV, a low-res representation is used as a placeholder during the time that the hi-res rep is being streamed into memory. If you wanted, you could force the lo-res rep to remain until is rendered into a certain region of the screen (e.g., the "center") at which time it switches to the hi-res version. This way, hi-res models aren't ever displayed unless they are directly in front of the viewer. Efficient? Probably. Beautiful? Probably not.

when the user sets the focus of his eyes at the screen border, the screen border will remain out of focus until the user commands the computer to move the camera. At the distance we use to watch the monitor it entire display is "almost" on focus.

It could work on a very big semi-spherical display (which I'm still waiting) though.

The game Myst 4 used something similar, they had depth-of-field effect that kept only the central part of the screen in focus, everything else was blured. Check out the Myst 4 demo and see it for yourself.

Needless to say, the vast majority of gamers promptly disabled that feature from the options menu and cursed the programmer whose idea was to put it in. It was also the source of many graphics drivers bugs.

Also, judging from the reactions on games like Gothic 3 and Call of Juarez, people really don't like this sort of bluring effects, they want everything to be crystally clear and pixel sharp.

I think its bad idea. The visual discontinuities aside, i doubt that you would really get any performance benfit fomr such a scheme. In my opinion it's more important to have an optimzed renderer than to use tricks such as this. Even shader LOD is often not very effective.