I must admit to being a little intimidated by the absolute erudition..
Erudition ? Nooooo!
If I may add my 0,02 euro to this discussion.
If the wavelength of light was infinitely small and if pure geometrical optics + classical lens aberrations were the ultimate model, and using a perfect detector with no noise and no resolution limit, I do not see any reason why a certain format would be better than another. Even classical DOF models would have to be re-visited wince DOF is not intrinsic to a given focal length or format, it is directly related also to a certain sharpness criterion. So beware, if you intend to do contact-prints-only with you Rolleiflex 6x6 TLR (like in most B&W family albums up to the sixties), the actual DOF might be much larger than what has been engraved on the focusing knob of your Rollei-TLR since... at least WW-II or even 1929
But in practice we have of course several limitations. In the age of film-only photography we had to live with the intrinsic limitations of silver halide layers. Well, in the past and still today, there has been gum-bichromate, cyanotypes, palladium-platinum, xerox copy, vidicon tubes and Zworykin's Iconoscope, and many other analog means of recording an image, but let's consider silver halide first since it has undergone the most of research an development for high quality still photography.
The advantage of film is that the performances of a given film in terms of sensitivity and grain/noise is basically unchanged when you coat one square metre of film for use in a mammoth still camera or when you use the few millimetre square of a super-8 movie camera frame. So definitely there is an advantage to use a big size of film with lesser final enlargement: getting rid of film grain and getting better gradations in gray levels because you have bigger film-pixels with little grain noise when scaling up the size of the medium. And since LF lenses are not zooms, designed for demanding professional users, and are always (or should be ) used at their best f-stop where diffraction balances residual aberrations, you get in LF the best optical quality that can be recorded in one pass. There is an on-going discussion about stitching, suffice to say that for copying flat objects, a scanner appears much better than a repro bench because the scanner lens does not have to transfer all the
image elements in one step, so the lens can be optimised for a smaller field and a better resolution, thanks to the stitching done in the scanner. So the good side of the demise of copying benches is the nice apo-ronars that we can get for affordable prices
But we canot forget that the wavelength of light is not infinitely small. Well 0.5 micron is not at our scale, so what's the matter ? In fact if you look at the best f-stop for 35mm, medium and large format cameras, you find that, very roughly, the optimum f-stop for standard lenses, those which cover about 50 to 70°, follows the empirical rule
f-best = (focal length in millimetres) / (8 millimetres).
For example the best f-stop for a 150mm covering 4x5 is about 16-22. the best f-stop for a 300 covering 8x10" would be closer to 32. The best f-stop for the 100mm Carl Zeis Hasselblad Planar is f-8. And the best f-stop for a 50mm standard lens for the 24x36 frame is about 5.6
Last generations of LF standard lenses like the apo-sironar or its competitor in Bad-Kreuznach probably exceed this figure, f/11-f/16 being the best f-stop and not f-16/f-22.
An amusing consequence of this scaling law is that if you relax the sharpness criterion proportionally to the format ( the CoC set to F/1720 or so) and if you always use the best f-stop for you lens, DOF is the same for all formats !! the only difference is that you'll need many more photons to get your LF image, and this is the price to pay for bigger film-pixels with a superb half-tones gradation and low grain noise.
If we extrapolate this rule to smaller formats, we can be frightened since the best f-stop for a standard lens covering the so-called 1/2" format, diagonal = 8 mm should be f/1 or so !!
Even if lens design and glass manufacture has made tremendous improvements, I do not believe in a f/1 lens covering 8mm in diagonal that would outperform, on the final print, the good ol' 75 mm tessar @f/11 on my Rollei TLR
The reason for this scaling rule is the balance between residual aberrations and diffraction. Residual aberratiosn scale like the format but diffraction does not. So all formats are not born equal, the larger, the better. Because of diffraction.
So to me there will always be ultimate advantage for larger formats, the fact that diffraction becomes negligible. Now which is the smallest format that is worth using? The border is really fuzzy and depends on how much you demand in terms of image quality and how good you detector is. And how much you ask for camera portabilty and compactness..
If you look back to the history of photography, the same story was told to both the professionals and the consumers : due to progress in film, you do not need this big camera. Stop using your Graflex, take a Rollei. Stop using your Rollei, take a Nikon. The story stopped some time in the '90s when even families realized how bad the quality of the 110 format was... and 35mm ruled. By charity, I'll not remember the story of the APS format. So why ultra-small digital formats are now considered as acceptable ?
The man advantage of digital sensors, now that we have a sufficient number of pixels is that the detection efficiency of silicon sensors far exceeds what best films can do. For example only the best films used for astrophotography could reach a few percents in terms of equivalent quantum efficiency, while current silicon photographic image detectors can reach about 20%. In other terms, you can use much smaller silicon pixels for the same final noise. Sure there are some issues like micro-lenses to collect photons, but anyway, silicon is intrinsically much better at catching photons that silver halide crystals are. We could remember that at the end of the XX-st century, a French team of scientists leaded by Mme Jacqueline Belloni patented a method to dramatically improve the efficiency of film.
http://www.cnrs.fr/cw/en/pres/compre...ionsphoto.html
The partner film company the the time was named... Agfa. No comment.
Film has a very low efficiency, so since Nicéphore Niépce Himself we have been waisting more than 99% of good photons. Improvements since the fist bitumen process (8 hours of exposure in bright sunshine) have been tremendous, and in the last 20 years of the XX-st century, we can say that 25-ISO film has been replaced by 100-iso films with better grain performance ! So probably a factor of 4 in efficiency or so. But still film cannot exceed a few percent in quantum detection efficiency.
Whereas a single silicon photodetector used in instrumentation can exceed 80% in terms of quantum efficiency. We are still far from this figure for tiny pixels in an image detector but there is hope for continued improvement. Whereas it is unlikely that developments following the discoveries made by Mme Belloni's team will ever been transferred to a real-world product...
So to me we have been lurred by considering lateral resolution and equivalent pixel count issues only. Sure, holographic films are capable of recording 5000 line pairs per millimetre and this is enjoyable for recording holograms, but what about detection efficiency?
To me and probably most of us the situation is that digital images do not require as many pixels as you would expect from the resolution limits of lens +film because they show virtually no grain noise in areas like skyes where there is no texture.
Two years from now, discussing with a French professional photographer who had been digitizing his LF slides for months and could compare to what he was directly recording with a digital SLR, he suggested that we should at least use twice as much surface of film to get the same final image quality after digitizing. Probably the real figure is something like 4 times in terms of equivalent surface, 2 times in linear dimension. You could object that you do not want to digitize film and wish to stay 100% analog in your darkroom, so the comparison might not be really meaningful. OK but behind this I can see only one reason why silicon images appear so good, i.e. the question of noise.
So a possible conclusion is that we could get the same image quality with a silicon sensor 4 times smaller in surface, and this is no mystery, a 645 digital sensor probably exceeds in quality what you'd get with a 6x9 digitized film. But anyway once you've made this factor 4 in surface or this factor 2 in linear dimension of the detector, the advantage of larger formats still holds due to the best f-stop issue, due to diffraction that does not scale. I think nobody will object that a 645 digital camera outperforms a four-thirds...
So when "they" will sell a 6x9 silicon sensor for less that 1500 dollars, sure I'll stop using film. In the past I had set to myself a limit of 1 megapixels before even starting to consider digital images. This limit was reached by the industry in consumer products much faster that I had expected. Even the cheapest camphone has now more pixels ;-)
So why don't I stop using film and switch to digital? Because even if I am convinced of the superiority of silicon over silver halide as an image detecteor, I am not a professional producing images for magazines, books or newspapers and making a living on ot. So I do not care for longer exposure times, slower process, expensive films (I'm spending more money in gas and car expenses to go on location for leisure landscape photography than I spend in film and chemicals) and I want to control all the process myself. Starting on the ground glass with my eyes. Not through an electronic display through a software.
Bookmarks