I was thinking about how Williams Tubes worked and how one could hypothetically "snapshot" (quite literally!) the state of a computer's memory by simply taking a photograph of the phosphor end of a memory-CRT - then feeding the stored state back into the machine by pointing a TV camera at the photograph and having the CRT display from the camera for a single refresh cycle before reconnecting back to the computer for state to be persisted on the CRT.
I imagine the technique could be used for any other computer that featured a visual display of its memory-state by way of der blinkenlights - this would assume that all its memory is visible this way. If a photograph of this was taken then a kind-of fax-machine could "scan" a photograph and use that to re-set memory would avoid the trouble of manually flipping switches to bootstrap a computer - I understand this is all within the realm of 1950s/1960s-era technology.
I know magnetic-tape and punch-cards predate computers and were used to store data (including bootstrapping) they share a considerable disadvantage: they cannot capture machine memory state in a "snapshot": they require memory to be streamed out first, thus freezing the machine's execution until all data had been written. They also suffer from relatively poor data density: a punch card requires about 2x5mm (10mm^2) area for each bit, while 1950s magnetic tape (e.g. UNISERVO) could store 128 bits/inch (0.19mm per bit) albiet in one dimension. Whereas photographic film is theoretically capable of much higher resolutions (thinking about microfiche, for example).
So, were photographs ever used to store memory state?