there will always be a limitation of precision, just as planck fundamentals are just used for convenience and may not actually exist. resolving power is a thing. it takes tons and tons of data to create an 'accurate' model of a protein, for instance, and it's still just a model, still just representative i.e. hydrophobic/philic regions, motifs, electronegativity, pKa, generalized structure - these are all values obtained w/ x-ray crystallography, MALDI, even experimental assay, and all this gets put together to make a pretty picture which is patently false but accurate enough to be useful.
actual pictures of a protein i.e. SEM/TEM or confocal are gonna be mostly useless and generally just show up in papers as "this is what it actually looks like" but it's more of a formality unless there's something specific that can be derived from the picture like fluorescence data w/ confocal. this is why confocal gets used a lot even though it's not as hi-res as SEM/TEM, which gives a very accurate picture which is for the most part of very low utility
