We've done this sort of things using a chain of shift registers (both with '595 and with dedicated shift-latch-current-source chips, in several different designs - 16x3xbi-color, 12x4 - I don't remember all of the setups). Large physical dimensions required certain clock shaping. Power distribution and mechanical setup was not trivial either... Yes, all driven from a vanilla '51.

If I recall correctly, some of them were multiplexed design, but I wouldn't do that again, especially since the shift-latch-current-source chips are relatively cheap nowadays.

JW