Driving a 2x16 LCD with only One Pin and 74HC595

[Darrel Taylor]

To shift out one byte (Data or Command) after LCDInit it needs 4 Steps.
One Step require a time nearly 910µs - 1ms.
That makes nearly 4ms. => No extra Delay necessary. Right?

No, each byte whether it's a command or data requires two nibbles at 910 uS or 1.82 mS total.

You need the delays!

And I would leave the initialization timing the same.
It only happens once per power-on, so you're not saving any time during normal operations.