Hacker News new | ask | show | jobs
by lb1lf 2936 days ago
When what you’re after is very high resolution, film still scales better than digital; the largest (somewhat common) film format is 8x10».

A sheet of slow (fine grain) film at this size can capture detail way beyond what any digital sensor I know of - surely it can be scanned with the equivalent of multi-gigapixel resolution - and if one were to develop such a monster sensor, it would have a hard time competing with the film in all but a few edge cases. (Most notably the ability to transmit imagery to the ground immediately, provided the necessary bandwidth is available.)
1 comments
libertyhouse 2935 days ago
Film alone doesn't determine performance - the whole instrument must be considered. The OBC trades "resolution" for area coverage rate. The digital electro-optical imagery from the U-2 is produced using an instrument (SYERS-2) with a different optical design. Intelligence Imagery (IMINT) is rated on the NIIRS scale and the OBC imagery is most-likely unclassified now due to it being at the mid-range of the NIIRS scale.

You make a good point about the scanning. Any wet film product these days must generally be scanned to allow for further production in our "soft-copy" world.

link
lb1lf 2935 days ago
-Film definitely is only part of the equation, and I have to admit I have no idea how optics scale (that is - is it easier or harder to make a lens with a large image circle with excellent resolving power, as opposed to a smaller image circle?)

My (attempted!) point was that fine-grain film* creates a negative which will be a challenge to match for most sensors out there - it is my understanding sensor yield drops drastically as it increases in size, so if you need a huge sensor, film is still the most cost-effective solution for a lot of applications. Then again, 'cost-effective' and 'national security' are two phrases which do not often come up in the same sentence!

*) My favourite - ADOX CMS20, has a resolving power which easily outperforms even a Summicron 50 at a sensible (not diffraction-restricted) aperture - more than 500lp/mm; the manufacturer claim up to 800lp/mm are resolvable (!)

That translates into some 12,500 line pairs/inch of film/sensor, as long as the optics are holding up their part of the bargain - or somewhere just short of 1 gigapixel from a 24x36mm frame, if scanned using -cough- sufficiently expensive equipment.

link