I like the VERA but it does feel a bit like it's taking on too much of the functionality of the Commander X16. It's write-and-internally-increment style of access seems to work really efficiently but I never did anything significant with the X16 emulator beyond some very basic BASIC programs.

I've taken quite a lot of inspiration from your VGA videos I hadn't seen (or taken in) that you were using '40103s but having just read your doc I saw the caveat about them potentially not being easily available. And they're twice the price of the '163. And their typical speed is 25Mhz vs the '163s 200Mhz. But it's half the chips and half the data lines. And I don't like doing subtraction (okay, pre-computing the negative widths is trivial and I suspect I'll really need my cycles on scanlines)

Yes, that was pre-VGA for me - I was doing composite/component output back then. Much lower frequencies! I have made some videos with that computer, but not for quite a while now.

In any case, I don't count individual pixels. The VGA timings for example are all multiple of 16, so for the sake of generating sync and blanking signals you can divide the clock by 16 before passing it to the 40103s. This is also how I get away with using 74HC590s in my VGA circuit - the clock is divided by 4 before it reaches them.