sorry for me looking so rude just trying to make a contructive discussion about future ideas
then then you'd either have more pixels (same total power) or 2 different charges (more pixels more IQ theoretical up to 380 NM maybe more if the momentum bounced somewhere else to a different position but lets stay for this 1)
lets say we have "2 types of lightrays or photons (particle wave duality)" 1 is hitting the sensor pixel no. 1 and was red
now comes a other photon from the object to the no. 2 pixel 5400 nm away and becomes a blue pixel
now the second scenario:
the object given away more lightrays(photons) "of the same color" and hit pixel 1 with the first lightray particle and the second lightray particle hits pixel 2
then you would have 2 of the same color or same total power in 2 pixels
in this case the charge would have the same charge for both 5400 nm sensor pixels and it would be 10800 nm that could be used
but thats an "if" it goes that way
not if both got a different charge
in case both get a different charge, then you lose IQ because the second information is lost
i only can speculate where the problem is, but the glass i have seen doesnt always produce a high amount of resolution
if the problem is elsewhere then you have to derminate how many resolution is produced and cancel the unneed´s pixels or combine the existing 1´s in a certain matter so this fits up for the IQ
you would have to know how many pixels you have to combine or not
but even if the most lead up to a pixel shift - if only a few pixels do not its a win/win for the IQ , more detail has gathered
the thing about the DDR4 bus was not about the y-sync speed it was about if the transfer of more data/charges from either the sensor or either the software is to much to downscale the pixels from the full sensor (according the pentax 645z is doing the same thing as the samsung sensor what takes 6 megapixels (and less size of is used to gather information) and doesnt downscale from a bigger amount of pixels
i didnt mean to see DDR4 as slot instead of the chips on it, those chips look like the attackment and can be put on a pcb
the thing why i mentioned DDR4 is that ddr4 also work like a rambus ram (rambus ram has own wires so it can be speaken or sended/readed to without having the bus busy)
then it opens a new question why the pentax cant record 24+ frames at 8256 * 6192 or at least 4096 * 2160 if the existing cpu and busspeed can already handle that
downscaling reduse noise in almost excellent way:
$L_{0}$ -Regularized Image Downscaling - Semantic Scholar Does downscaling a picture remove noise? : photography noise - Downscaling by skipping rows and columns ? any example images? - Computer Graphics Stack Exchange
when you took a picture in less light with 400*400 pixels you got a lot of noise
but if you took that picture in 4000*4000 pixels and downscale the information what is left or present in the 4000*4000 picture "almost" fix up that noise
the full sensor should be readed out and downscaled, not what i suspect it just use less of the sensor and make the pictures out of that
its a super win/win here
thinking about a codec i could think about the new h265 , with options for "deinterlacing" and a "user setable Bitrate"
the h264 also could be as second option
sorry if i offended any1
here is a picture of a ramchip (not using a slot or something) how it could look on a pentax camera pcb