I’m well-aware if that.You are ignoring other reasons why one might change sub-pixel geometry that have nothing to do with color efficiency. The driving circuits for each sub-pixel need to go somewhere and are not symmetrical in layout for all 4 colors. This is why you get those weird shapes instead of simple rectangles of various sizes for each color. They have buried circuit traces and components inside some of those indentations. The spare room varies by pixel pitch which is why the shapes change with panel size. Obviously color efficiency plays a role too but it needs to be balanced with what they can fit in the available "dead space" of the panel.
In the end, the pixel full-factor depends on the ‘dead space’ needed for subpixel circuitry, the minimum inter-subpixel distances required, and the dead space caused by any top traces routed in non-transparent metal (is that what causes the large black lines between WOLED rows?).
But no matter the pixel size and no matter how much dead space is needed for curcuitry, in the end it all translates to relative mm^2 (or rather um^2) of W, R, G and B subpixels.
Subpixel size determines current density and lifetime / aging and the only reason you’d choose a different set of active subpixel area ratios is if you decided to match performance / specifications of a larger panel with larger fill factor to the performance of a smaller panel with smaller fill factor (meaning you’ve got excess real-estate / subpixel area to burn).
I circled back and had another look at the 88Z9 subpixel shot by HDTVTEST, since that’s the smallest subpixel we’ve seen (4K @ 44” equivalent). The 88Z9 has a larger white subpixel relative to the colored subpixel than any of the other designs we’ve seen. I’d characterize it as G < B < R << W, and this should mean higher peak whites but at the expense of lower peak fully-saturated colors.
So comparing 2019 WOLED sub-pixels for different panel sizes, we’ve got:
88Z9: G < B < R << W
55C9: G = B < R < W
77C9: G < B < R < W
(and 83C1: G < B < R = W)
Obviously, this is very handwavy and wend need someone with the wherewithal to translate these pictures and eyeball estimates into measured % for each subpixel to confirm this trend definitively.
But that being said, to me it seems likely that LGD has made the decision to maintain peak white levels across different panel sizes of a generation and to use additional available pixel space on larger panel sizes to increase peak fully-saturated color levels (or just to increase lifetime if they’ve decided to fully-match color volume, even at the fully-saturated extremes).
Geometry / design changes to accommodate subpixel circuitry, but overall area does not (need to, at least)…