Lehti: So how does that differ from supersampling?

Bluekkis: Supersampling renders the scene once in higher resolution and then scales it down to screen resolution.

ragholio: Ubersampling behaves more like the renderer in proffesional 3D software such as 3DS MAX and Maya. The frame or element is sampled multiple times (not sure how many in this game but probably between 4 and 8) and then jitter composited. This results in much more accurate rendering of all screen elements such a shadows, SSAO and transparencies. For example, if the frame is only rendered once with 8 samples per pixel then the effects show artifacts (as can be seen with Ubersampling off and sharpening on), but if that frame is rendered 8 times then the image will be smoother and these artifacts will become less prominent. Likewise a frame with 16 samples per pixel will only need to be rendered 4 times to achieve the same result. I'm guessing that the game uses between 4 and 8 samples per pixel, per frame as 16, 32 or higher would be far too costly in terms of rendering overhead.

Note: I'm not talking about frame bit depth here, wich is most probably 32.

Confused? Don't worry. The important thing to know is that you should turn it off until about 2013 when the next wave of consoles hit and the average PC has enough grunt due to it becoming a commonplace effect.

kendrix: well, there is a difference. with the setting on, the game gets unnecessary blurred. which sucks, because tw2 has some great hires textures.

Lehti: Ok, I think I got it. Is there a more common term for the algorithm other than "übersampling"?

predcon: No, that's the "motion blur". Apparently, from the descriptions I'm reading, Ubersampling is rendering each frame multiple times, which exaggerates a function that already exaggerates a visual device meant to convey the "blurring" the human eye perceives as when turning your head or moving your eyes really fast. Imagine watching a DVD on "Single-frame-advance" mode during such a time as when the motion blur function is being displayed. It only looks like it's more blurry because Ubersampling is slowing down the framerate enough to keep the blur effects on screen longer than was intended.

predcon: Probably not, since it looks to be a proprietary CDPR technology. Their tech, their terminology.

predcon: Probably not, since it looks to be a proprietary CDPR technology. Their tech, their terminology.

Lehti: Ok, I think I got it. Is there a more common term for the algorithm other than "übersampling"?

cjrgreen: It used to be called "full-screen anti-aliasing" (FSAA). It was dropped by GPU makers because it is so inefficient, then reinstated in recent GPUs because they have enough computing power to do it (Fermi and Northern Islands GPUs have it).

It's pretty inefficient to compute, because you need to render every frame 2 to 4 times, rather than traditional AA, which applies a more capable filter only to the edges in the frame designated for AA. Also, its performance is that of a simple first-order moving-average filter, which does little except soften the image's focus.

Lehti: Unless I missed the point by a mile, Ubersampling is an advanced and computationally intensive (eats FPS) anti-aliasing technique. Doesn't contribute to motion blur at all.