A few comments as yours is close and I’m too lazy to do the write up myself, bumming off your work:

• Nyquist frequency is half the sample, so your effective pixel for visible light will actually be at 380nm / 2
• An electron microscope can capture down to 2 nm so actually 2 PB is a limiting factor for that! 29.6nm would definitely be possible.
• Image compression ratio (assuming PNG) would probably be 2-4, so the actual pixel size would be sqrt(29.6) to 29.6 / 2.