Blur Busters Forums

Chief Blur Buster wrote:Wow --

Even at this early juncture, this article is apparently getting rave reviews by Blur Busters twitterverse, including employees of NVIDIA and media websites (e.g. Digital Foundry). I try to explain things better than the average website!

• Using sample-and-hold method for static parts of the image

• Using strobed method for eye-tracked moving parts

• Using sample-and-hold with introduced motion blur effect for any residual motion that is not eye-tracked

punpcklbw wrote: ↑

09 Feb 2025, 19:37

Yes, if you get rid of motion blur, you'll have to deal with the stroboscopic effect instead. It would take groundbreaking advances in the display technology to transcend finite framerates, comparable to the transition from early vector displays to today's raster displays.

The bandwidth requirements for a 1000+ Hz high-definition video signal would be immense. But do you actually need framerates this high? I suppose that double the flicker fusion threshold should be sufficient (what evaluates to the 140-180 Hz range being the sweet spot). The key is to split the signal into separate sample-and-hold and strobed components and display both. That wouldn't work for LCD and any back-lit panels, neither for CRT due to the lack of sample-and-hold ability. We're only left with advanced LED displays, where you can achieve lifelike (or close to that) motion by:

• Using sample-and-hold method for static parts of the image

• Using strobed method for eye-tracked moving parts

• Using sample-and-hold with introduced motion blur effect for any residual motion that is not eye-tracked

This way, you only have to distinguish what is eye-tracked by the user and what is not, requiring some realtime user eye tracking. But at least you wouldn't need crazy ass refresh rates to make it work. And even without any eye tracking this could be a step forward in image clarity and liveliness.