Discussion about Phosphor Fade Simulation / Rolling-Scan Software BFI

[Chief Blur Buster]

Simulate CRT 60Hz phosphor decay in LCD strobe 60Hz is good for less flicker, but undesirable for motion quality, panning in CRT 60Hz a white moving object in a fullscreen black background, cause a long trail WhiteToBlack behind the white object. This same test in XG270 SingleStrobe 60Hz with strobe fast WhiteToBlack, for sure leave much less trail, but with heavy blink

The solution can be set a long strobe lenght at 60Hz for compensate, trade Flicker with MotionBlur.

We'd need 1000 Hz LCDs to more accurately emulate phosphor decay -- since phosphor decay can't be emulated accurately at 16.7ms granularity at 60Hz (stepped fade effect with LCD-like ghosting effects). Phosphor decay ideally needs to be emulated at millisecond, even sub-millisecond granularity. One could implement a software-based rolling-scan BFI, e.g. theoretical 4-segment BFI for 60fps at 240Hz with gamma-corrected alpha-blended zones between segments, an equivalent of phosphor decay emulation.

[AddictFPS]

We'd need 1000 Hz LCDs to more accurately emulate phosphor decay -- since phosphor decay can't be emulated accurately at 16.7ms granularity at 60Hz (stepped fade effect with LCD-like ghosting effects). Phosphor decay ideally needs to be emulated at millisecond, even sub-millisecond granularity. One could implement a software-based rolling-scan BFI, e.g. theoretical 4-segment BFI for 60fps at 240Hz with gamma-corrected alpha-blended zones between segments, an equivalent of phosphor decay emulation.

For 100% emulation is the only way, all screen pixels as a different decay "start time" and "end time". But currently, the XG270 backlight can be programmed to make the strobe lenght, Off-On close to instantaneous, and On-Off a progressive fullscreen decay ? a OSD "Under 75Hz mode" setting to apply this behavior with current PureXP+ strobe lenght levels.

Scanout speed 240Hz 4.16ms -> all pixels have the signal to change color
GTG máx time (100% transition) +9ms (like ACER XV273X with same panel) -> at 13.16ms all pixels has the stable color
From 13.16 to 16.66, the are 3.5ms to play with this decay without crosstalk, it could work ?

You'll still have to rely on software BFI + 120 Hz in your favourite emulators.

Anyone know software BFI that work with GPU drivers, in order to make BFI+120Hz in DirectX OpenGL and Vulkan games capped at 60FPS ?

[Chief Blur Buster]

Fade emulation on strobe backlights will only go part way.

The problem with flicker perception is that strobing is global, while CRT is rolling scan.

CRTs feel like they flicker less because something is being illuminated somewhere all the time.

If you see a high speed video of a CRT, you will see a phosphor dot or bar at all times.
The total amount of light hitting human eyes remain relatively constant over the milliseconds

If you see a high speed video of a strobe backlight, you will see some moments that a screen is completely black.
The total amount of light hitting human eyes varies wildly over the milliseconds

This cannot be emulated very easily with sample-and-hold displays at current refresh rates, unless there's an expensive FALD scanning backlight. See this thread -- Scanning Backlight Via a FALD

[Chief Blur Buster]

I have opened discussion threads at a few emulator forums, to highlight this potential, as a "Temporal HLSL":

http://forum.arcadecontrols.com/index.p ... 926.0.html

https://github.com/mamedev/mame/issues/6762

This is probably a long-term incubation, but I will probably create a TestUFO CRT Emulator in the coming years, as it's relatively simple for me to create.

I'd likely make it a Hz-agnostic CRT emulator, and use gamma-corrected alphablend on the overlapping areas of the rolling-scan segments.