Originally posted by chicagonature For Everyone,
Can anyone explain some of the reasons why a lens is better with film than it is with a digital sensor? I understand that shiny sensors cause light to bounce around with the lens, but I'm talking more about sharpness and resolution. Also, remember that the 645 lenses are true 6x4.5 lens, whereas the 645D sensor is smaller and is able to take advantage of the lenses sweet spot. So, what's up with that?
Mike - there are *many* factors at play in digital versus film, and that's quite a topic to go on about in and of itself. I'm not a real techno guru but I'll give it a go for starters ...
Technical hooey - A few factors you may want to read about concerning CCD/CMOS sensors vs. film, along with my unqualified explanation the way I understand it.
- Bayer filtering - RGB is sensed in 2x2 horiz. arrays (1R,1B,2G) <except Foveon sensors>; whereas film is 3 vertical layers/substrates. The upshot to this is photons exciting a CCD may or may not hit the right color photodiode on the array (again due to many reasons like channel width, microlenses, etc.), note the pixel size is quite small (e.g. 5 micron). In film however, where the grain size is ~10 microns, generally speaking a single photon of a particular wavelength nearly always will strike the correct color layer (as the grains are vertically aligned).
- CCD microlenses - these are small lenses atop each pixel to increase the amount of photons hitting the sensor buckets. Suffice it to say they don't always target the right photodiode for a given wavelength.
- Analog lens designs are meant to gather light of all wavelengths and deliver them to the film plane at a wide range of angular incidence, whereas CCD/CMOS sensors (due to the walls surrounding the buckets) can't tolerate extreme (high) angles of incidence (more like they want it perpendicular). The end result is the CCD/CMOS sensor gets less photons to eat; hence the Signal to Noise (SNR) ratio is lower than would be for a given 10 micron section of film (if using an analog lens).
- Because of their smaller size (format and "grain" or pixel size) and their nature CCD/CMOS sensors don't react / fare the same as film in regards to the effective aperture of the lens. Essentially there's a lot of stuff getting bounced around/absorbed, hence F22 on film may look great - on digital it will result in really soft images.
There's a ton of resources on the web about this btw.
On the more practical advice side, I've worked as an amateur with MF for about 20 years. I gave up film about 9 years ago. I currently have a mixed camera arsenal of Mamiya RZ (every lens), PhaseOne 645 (a wide range of mostly Mamiya lenses), Contax 645 kit (except the 350 APO), Nikon kit, and even odd looking technical cameras. MF digital capture is on Phase P65+ and Leaf. All of that in search of replacing my fond film days - and I've never achieved that "look" or romance (but hey I'm still trying).
For the MF lenses I've come to a few general but very subjective conclusions about my some odd 50 analog MF lenses on digital using Kodak/Phase/Leaf:
- Most *MF* lenses (except Zeiss) were mass produced with mediocre standards in the last 30 years (pre digital). Thus you get a huge variation in the quality for a given lens.
- Bigger glass is always sharper and more contrasty (e.g. I'd prefer an RZ lens any day over 645) - I think this is because the exit pupil is bigger and the sensor gets to eat more perpendicular photons (take a look at a 67 lense at f11 vs. a 645).
- Primes are always sharper
- Faster isn't better unless you're in low available light.
- Zooms (including Zeiss) - plainly suck (but they're convenient!)
- Don't go wide open unless its night time and your shooting bums.
- Don't ever do anything tighter than f11 unless its a pinhole
- AF on any MF camera (up to and including the 645D and Phase/Mamiya 645DF, and Leica S2 or Hassleblahs) blows chunks, even my D1 of 11 years ago had much better AF.
Having stated the above it might sound as if I'm saying analog lenses used with digital backs aren't useful for good photography. I think the issue is I (we?) tend to over-analyze the digital output whereas I didn't have a 250x loupe handy 10 years ago to know the difference.
Sure these lenses have shortcomings on digital - and in 5 years time a new breed of Pentax digital lenses will likely replace the current AF analog line available now - but in the meantime we have to make do with what we have. That's not true for some other pricey makes with Zeiss or Schneider glass but the entry point for those systems is in the $50K range. So we have to adapt and innovate our techniques unless we're ready to spend that kind of cash.
Here's an example for you how to adapt your analog lenses - in your original post you stated you shoot mostly f/16~f/22 deep view landscapes with closest focus in the range of a couple of feet to the horizon. You ain't gonna do that sharply in digital at f/16 or f/22. So .... why not shoot (digital) focus bracketed at f8 (or the sweet spot for a particular lens) with maximum sharpness and then stack the frames in post processing ? The resulting print would be sharper, deeper, and more contrasty than you could ever achieve at f16 or 22 on a single frame of film (or digital). Note you can do this with scanned film as well (ain't that convenient thou). There's a famous landscape photographer out there using that technique now on a P65+ with an 8x10 Deardorff for sale btw, and I bet he didn't look twice (nor knew how to do so initially) at a lens resolution test or endless frames of brick walls.
Sorry for the long post - its cold out tonight and I'm too awake anticipating my 645D arriving tomorrow