Blur Busters Forums

16kHz



RoadToVR article: http://www.roadtovr.com/nvidia-demonstr ... r-display/

Fantastic!

As much as the general population this is unnecessary -- I already knew we'd have to get to the 10KHz refresh rate leagues eventually to pass the theoretical "Holodeck Turing Test" (basically, the theoretical "Wow, I didn't know I was wearing a VR headset instead of ski goggles" where VR looks exactly like real life)...

To eliminate strobing as a form of motion blur reduction, we really need really high frequencies to avoid wagonwheel effects / stroboscopic effects. Finite refresh rates (the artificial human invention of using a series static images to approximate motion) always
has indirect side effects that affects ability to reach Holodeck perfection.

Humans detecting side effects from 5000Hz and above:

(from a lighting industry paper -- but also applicable to strobe-backlight displays).

At 16KHz refresh rate, we would not need strobing anymore as a means of motion blur reduction. Full-persistence frames of 63 microseconds each (1/16000sec) would be fantastic, even if DLP limitations still apply here (e.g. temporal artifacts like rainbows) and still need to be fixed later in technological progress...

In certain situations -- even at a 5000Hz refresh rate -- humans can still indirectly see side effects of a non-infinite refresh rate that distinguishes from reality (as reality is not consisted of a consecutive series of static images). Some of the theoretical discussion is in So what refresh rate do I need?, discusses when the points of diminishing returns virtually disappears -- and there are certain use cases (VR) where it doesn't happen till we reach 5-digit refresh rates...

This requires a lot of interpolation creativity, and this is only for AR (and currently, monochrome 1-bit DLP).

That said, If we were able to do full color at true full 16KHz end-to-end, it would also completely eliminate mouse dropping effects too -- even on 4K displays -- no matter how fast the mouse cursor moved, it would be a continuous motion blur (cursor position updates every pixel) rather than dotted / phantom array. You'd be able to fixed-gaze (e.g. a crosshairs) and the background would scroll seamlessly like real life -- without stepping effects / phantom array effects -- of a finite Hz.

Chief Blur Buster wrote: It would also completely eliminate mouse dropping effects too -- no matter how fast the mouse cursor moved, it would be a continuous motion blur (cursor position updates every pixel) rather than dotted / phantom array. You'd be able to fixed-gaze (e.g. a crosshairs) and the background would scroll seamlessly like real life -- without stepping effects / phantom array effects -- of a finite Hz.

That would be my dream display.

I don't think it will help at all. GPUs are not capable of such framerates.

RealNC wrote:I don't think it will help at all. GPUs are not capable of such framerates.

My favorite multiplayer game was (I stopped playing it) a 2D shooter that I could run at several thousands frames per second.

We need to signal to game developers that performance matters and that the future is high frame rates.

RealNC wrote:I don't think it will help at all. GPUs are not capable of such framerates.

The cool thing about that video though is they are using a 60Hz source and some kind of frame interpolation plus pixel transition offset algorithm to make the motion almost perfectly smooth. That's 1/8th playback speed, so emulating 480Hz. Pretty neat!

cskippy wrote:

RealNC wrote:I don't think it will help at all. GPUs are not capable of such framerates.

The cool thing about that video though is they are using a 60Hz source and some kind of frame interpolation plus pixel transition offset algorithm to make the motion almost perfectly smooth. That's 1/8th playback speed, so emulating 480Hz. Pretty neat!

That's AR (augmented reality). AR is about projecting an image (computer generated) on top of another one (real world.)

Higher refresh rates provide huge benefits for AR. For non-AR use, the benefits are rather small. In fact, the only benefit I can think of when running, say, 150FPS@16kHZ is that you don't need to care about vsync or VRR anymore. Animation is going to be judder-free regardless of frame rate. (Judder is still there, but at 16kHz it's undetectable.)

But, you know, VRR is the actually practical solution to the problem

RealNC wrote:I don't think it will help at all. GPUs are not capable of such framerates.

I would choose 2004 era graphics at ultra high refresh rates over the current situation. At that point I think the bottleneck would be on the CPU side.

Dew wrote:

RealNC wrote:I don't think it will help at all. GPUs are not capable of such framerates.

I would choose 2004 era graphics at ultra high refresh rates over the current situation. At that point I think the bottleneck would be on the CPU side.

Unfortunately, even very old games do not run at such frame rates. They utilize only small parts of the GPU (mainly geometry.) So what happens is that the GPU is usually idle.

However, 1000Hz, or even 2000Hz is realistic for old games. 16000Hz... eh, not so much. A 1000Hz OLED display is (IMO) what we can realistically dream for.

RealNC wrote:However, 1000Hz, or even 2000Hz is realistic for old games. 16000Hz... eh, not so much. A 1000Hz OLED display is (IMO) what we can realistically dream for.

Now we just need 8-channel 120Hz OLED panels:



(From Custom OLED Rolling Scans thread).

Overlapped concurrent scans may make high-Hz more doable within our lifetimes. Each pixel is still updated exactly 1/960sec apart, and retains the full temporal resolution of 960Hz while being achieved using a much slower full-panel 1/120sec scanout velocity. Frame delivery speed into the display will be another issue, however, 8K 60Hz is the same bandwidth as 1080p 960Hz.