IMO the key insight of M4 / AM4 could easily be applied to a wide variety of cases where you have way more data than pixels. Making "TableLens but for modern datasets" is possible with these kinds of techniques.
finally used that M4-like alg to do pixel-binned column values (far right side). These sparks show the actual value of the column as a function of row number
akin to one of the core concepts of TableLens, but this viz will scale to arbitrarily large table sizes for fixed pixels