Dumping the NEC PC Engine YUV Table

Very very cool and ambitious, it's always instructive (and somewhat scary) to see people mixing looking at the actual integrated circuit with low-level programming and spaghetti-style wiring to get information about.

Anyway, not sure I understood the issue right, but shouldn't it be possible to at least get a decent approximation of the suspected color-mapping by just writing a program to display all 512 possible colors (it's 3-bit RGB) on screen, digitizing 1,000 frames of "pure" RGB output along with 1,000 frames of YUV, and comparing?

This is some top tier hacking and reverse engineering, incredible stuff.