Page 1 of 1

Is 1000Hz+ interpolation of 200Hz native content enough?

PostPosted: 15 Jan 2016, 18:48
by AustinClark
Is 1000Hz+ interpolation of 200Hz native content a realistic and practical way to eliminate the stroboscopic effect (as well as persistence problems)?
I figure any artifacts created by interpolation will be largely unnoticeable at 200Hz, and that persistence of vision solutions (such as backlight strobing) won't even be necessary. In games, you would have to choose to either use predictive interpolation or buffer an extra frame. At 200Hz, that should be fine for most people and most games (100Hz effective latency, at best). OLED displays would be necessary, due to the low response times needed, but the bandwidth needed is definitely possible, especially with Display Port 1.3 (at anything under 4K) and beyond.

My main concerns would be processing the interpolating frames, and physically driving the pixels at that speed. Integrating a system like G-Sync would also be problematic.

Thoughts?

Re: Is 1000Hz+ interpolation of 200Hz native content enough?

PostPosted: 16 Jan 2016, 04:00
by RealNC
The stroboscopic effect approaches zero as the refresh rate approaches infinity. So to get rid of it, you would need a monitor with an infinite refresh rate :mrgreen: However, 1000Hz would eliminate the effect for motion speeds of 1000 pixels per second or less. Which is "decent" ;-)

(For the effect to be eliminated completely, all pixels in motion must pass through all monitor pixels between the start end ending points without skipping any of them. At 1000Hz, everything up to 1000 pixels per second can do that.)

Re: Is 1000Hz+ interpolation of 200Hz native content enough?

PostPosted: 16 Jan 2016, 08:06
by lexlazootin
all pixels in motion must pass through all monitor pixels between the start end ending points without skipping any of them. At 1000Hz, everything up to 1000 pixels per second can do that.


that's clever, I've never thought of it like that. I imagine persistence would be pretty sweet at 1000hz aswell.

I think it's crazy that they haven't released 300hz monitors yet. is it really that hard?

Re: Is 1000Hz+ interpolation of 200Hz native content enough?

PostPosted: 16 Jan 2016, 11:55
by spacediver
lexlazootin wrote:I think it's crazy that they haven't released 300hz monitors yet. is it really that hard?


As refresh rates get higher, you need faster pixel rise and fall times to avoid cross talk between refreshes. Also keep in mind the increased video bandwidth requirements to drive high resolutions at higher refresh rates.

Re: Is 1000Hz+ interpolation of 200Hz native content enough?

PostPosted: 16 Jan 2016, 21:48
by lexlazootin
spacediver wrote:As refresh rates get higher, you need faster pixel rise and fall times to avoid cross talk between refreshes.


Doesn't that more or less just happen as the refresh gets higher? http://i.imgur.com/NEpjSTM.png

I thought DP3 can do UHD at 120hz? beep bap boop, that's 1080p at 480hz no?

Re: Is 1000Hz+ interpolation of 200Hz native content enough?

PostPosted: 16 Jan 2016, 23:53
by RealNC
He means the panels themselves. If a pixel cannot be updated that many times per second, higher refresh will give you washed-out colors and blur. LCDs need time to change state.

This is already happening with 144Hz monitors. If you actually measure color reproduction, the colors at 144Hz are worse than at 60Hz. Not very much so, but it would get even worse if you up the refresh rate even more. For the pixel to fully transition into its final state, it needs time. If it changes state before it has transitioned completely, you're getting a wrong color.

Re: Is 1000Hz+ interpolation of 200Hz native content enough?

PostPosted: 19 Jan 2016, 20:18
by AustinClark
RealNC wrote:He means the panels themselves. If a pixel cannot be updated that many times per second, higher refresh will give you washed-out colors and blur. LCDs need time to change state.

This is already happening with 144Hz monitors. If you actually measure color reproduction, the colors at 144Hz are worse than at 60Hz. Not very much so, but it would get even worse if you up the refresh rate even more. For the pixel to fully transition into its final state, it needs time. If it changes state before it has transitioned completely, you're getting a wrong color.


Yes. In fact, I have a QNIX 1080p TN panel that will overclock all the way up to 240Hz, but it looks absolutely terrible. Even at 144Hz, Gamma is WAY too high. Trying to compensate with windows calibration tool doesn't cut it.

That's what got me thinking "Is this enough?". Aside from stroboscopic effects (and minimal persistence of vision blur), going over 200Hz seemed to provide little improvement (Although I think I may have been hitting a pixel response time bottleneck. These monitors don't have any overdrive, which is one of the reasons they can be successfully overclocked). I figure most people would prefer higher resolutions and graphics settings to improved framerate well before 200+Hz was met anyway. Smoothness was quite good, persistence of vision blur wasn't THAT bad (and could still be improved with backlight strobing anyway), and input lag was probably about the lowest I've experienced (but I haven't done tests). It's not PERFECT, but considering the trade-offs involved in generating anything higher (Performance and bandwidth requirements), 200+Hz is about the realistic limit. We'll have to find other ways to improve, hence my interpolation idea. VR might be a different story.

Personally I will sacrifice graphics settings to get at least around 80-100Hz stable (60 isn't my target, it's my absolute minimum) so I'm pretty picky.

Too bad there isn't a way for me to directly compare to a "perfect" monitor. That would provide the insight I need.