'Solving Transparent Z-Buffering

[testkonto-kn]

when the screen resolution is higher so a single projected pixel would fill several pixels on the screen, the likelihood of each of the several pixels drawn or set on the screen and also z buffer for one point of the terrain could be randomized according to the transparency and afterwards one could blur or smoothen the screen so statistically with the gaps it would blend the pixels together even?!?

[testkonto-kn]

z buffer for each pixel most in front before the rotation and central projection for each screen coordinate, well, for transparency, say, having four z buffer values and sorting each pixel in that array for the coordinate before drawing according to the distance also with the transparency information, if each pixel covers up with at least 25 per cent there, would this be correct to just weight the average value from back to nearest subsequentially and have the z buffer only four-fold(has to start with some background color), just would require to scale down the number of pixels projected proportionally(?) to the z distance, but this is coarse a calculation time advantage also, i am considering pixel clouds, sorting and calculating blended values could also be parallelized so speed is still without theoretical limit?!?