Strobing-Aware ABL
Posted: 08 Mar 2023, 01:00
I wanted to make sure you know about a concept that I and some of the folks in my discord server figured out.
Strobing-aware ABL:
Normally, OLED monitors don't adjust their ABL curve for when strobing is active. A full screen white window on the LG CX for example is 146 nits. If you use software BFI of 50% duty cycle then the luminance of that same window size will be halved to 73 nits. This is not optimal. The ABL isn't aware of pixels being turned off. If the ABL curve were designed to be aware of strobing duty cycles then the 100% window during 50% duty cycle strobing will be half of 324 nits(the unstrobed 50% window luminance of the LG CX), which is 162 nits. The pixels are not in this case being boosted! Voltage wise they are rated to be this bright, they are not going outside of spec as far as that goes. Time wise they are not getting any more wear than if a 50% window was up during sample-and-hold. Heat wise all the pixels being lit for half the time(at the same drive) is not any different to half the pixels being lit(at the same drive) for all the time. Extend this logic to a tiny window size and a tiny duty cycle(say 10%) in HDR mode and we have 804 nits on sample and hold on the CX. This when multiplied by 0.10 is of course 80 nits. Very manageable! It's easy to visualize when you imagine the scanning bar as a percentage of the display area. There is no difference heat/voltage/wear wise between 10% of the screen lit 100% of the time to 804 nits in sample-and-hold mode and 100% of the screen lit to 804 nits 10% of the time in strobed mode. We just have to make the ABL aware of that.
A parallel example to this is scanline shaders. ABL is already aware of scanline shaders. Everything I have said so far is already in action when using scanline shaders on oleds, I have tested it with my oled pvm-2551md. 50% scanline shader on full white screen = 50% window luminance divided by 2, rather than 100% window luminance divided by 2.
Source for the luminance of CX: https://www.rtings.com/tv/reviews/lg/cx-oled#test_180 the SDR brightness section.
Strobing-aware ABL:
Normally, OLED monitors don't adjust their ABL curve for when strobing is active. A full screen white window on the LG CX for example is 146 nits. If you use software BFI of 50% duty cycle then the luminance of that same window size will be halved to 73 nits. This is not optimal. The ABL isn't aware of pixels being turned off. If the ABL curve were designed to be aware of strobing duty cycles then the 100% window during 50% duty cycle strobing will be half of 324 nits(the unstrobed 50% window luminance of the LG CX), which is 162 nits. The pixels are not in this case being boosted! Voltage wise they are rated to be this bright, they are not going outside of spec as far as that goes. Time wise they are not getting any more wear than if a 50% window was up during sample-and-hold. Heat wise all the pixels being lit for half the time(at the same drive) is not any different to half the pixels being lit(at the same drive) for all the time. Extend this logic to a tiny window size and a tiny duty cycle(say 10%) in HDR mode and we have 804 nits on sample and hold on the CX. This when multiplied by 0.10 is of course 80 nits. Very manageable! It's easy to visualize when you imagine the scanning bar as a percentage of the display area. There is no difference heat/voltage/wear wise between 10% of the screen lit 100% of the time to 804 nits in sample-and-hold mode and 100% of the screen lit to 804 nits 10% of the time in strobed mode. We just have to make the ABL aware of that.
A parallel example to this is scanline shaders. ABL is already aware of scanline shaders. Everything I have said so far is already in action when using scanline shaders on oleds, I have tested it with my oled pvm-2551md. 50% scanline shader on full white screen = 50% window luminance divided by 2, rather than 100% window luminance divided by 2.
Source for the luminance of CX: https://www.rtings.com/tv/reviews/lg/cx-oled#test_180 the SDR brightness section.