If all you're trying to preserve is text, then a digital format would make sense. Just add "analog" images of unicode tables covering all used symbols, and any

If all you're trying to preserve is text, then a digital format would make sense. Just add "analog" images of unicode tables covering all used symbols, and any used error correction format. But if this is trying to entice libraries with "replace your microfiche with this easier to maintain format, and do something for history while you are at it" then you need to store scans of newspapers, magazines, etc. Which probably means storing PNGs or similar. At that point, I'm not sure the capacity improvement over "analog" optically readable images is really worth the much more complex decoding setup. Both for the library use case and for the "a civilization that might not speak our language wants to decode this" use case the complexity of the decoding setup is a major factor.