3 ms·

There is a multitude of practical reasons. First of all, modern digital sensors offer resolutions far beyond what normal film material would offer. So 35mm came

by _ph_ 5y ago

There is a multitude of practical reasons. First of all, modern digital sensors offer resolutions far beyond what normal film material would offer. So 35mm cameras easily beat the resolution of classical medium format film. You need lenses to match that resolution. Then there are new requirements like AF and optical image stabilisation. With classical prime lenses, focussing often would move the whole lens forward and backward. Or at least some large lens group. That makes the optical construction easier but prevents fast AF. Modern lens designs would use a rather small lens in the middle for focussing. With the reduced mass, the focus speed goes up. Image stabilization also asks for a small lens or lens group to be moved. But one big difference shouldn't be overlooked either: the abundance of compute power. Today, you can easily simulate a lens system of any complexity. Leica did actually use raytracing to optimize their lenses in the 50ies - but in the absence of computers, they had a room full of people who did the calculations manually. So with a months worth of work, they managed to calculate the path of like 5 rays through their given lens design. Which had like 6 elements. Modern prime lenses can have up to 20 lens elements, zoom lenses up to 30. For a good comparison about different lens design approaches, one should look at Leica lenses. Because Leica both still makes classical manual focus lenses for the M system, which typically have like 5-7 lens elements, and absolutely modern style lenses for their SL system, which have twice or more lens elements. In each design philosophy, they are making some of the best lenses, cost usually not the top priority :). Which makes a good base for comparisons. For example, they quite recently both introduced a new 35f2 lens for the M and the SL mount. Personally, I am quite fond of the prime lenses with few lens elements. While modern glass coatings can pretty much eliminate reflections on glass surfaces, every single glass-air boundary does create reflections. So while the modern designs are technically close to perfect, I think those designs with few lens elements have a special image rendering. Very contrasty, lots of character.

klodolph 5y ago
As someone who sometimes shoots medium format film, sometimes shoots digital—I am a bit skeptical about claims that digital “easily beat the resolution of classical medium format film”. I shoot with a nice, but old, MF camera and prime lens on a tripod and then do wet prints (usually) or drum scans (rarely). Based on close examination of drum scans, I estimate my setup gives images of around 60-80 megapixels. Knock that down somewhat for color. This is a bit lower than the quoted resolution for the film I use, if you read the spec sheet. Rough comparison of digital vs film leads me to conclude that you can’t just compare the camera resolution against the film scan resolution. Maybe it’s the low-pass filter / Bayer filter, but a film scan will have the same image quality as a raw digital photo of somewhat larger size (say, 50%). So if you take a high-end full-frame digital camera with a 24x36mm sensor, you get 50 megapixels, and my film setup will get you around the same number of pixels when shooting color, but the film scans will be sharper. The film is 4x the area—56x70mm. So it’s close, and I wouldn’t be surprised if a top-of-the-line full-frame digital camera could beat a 645. But 645 is the smallest of the medium format sizes, and digital wouldn’t be beating it by a huge margin. It’s a bit subjective, because the transfer functions for digital and film are so different and there’s no obvious correct way to compare them mathematically. That said, digital will easily win at higher ISOs.
- leemailll 5y ago
  I suspect when people compare digital against film, they often mean the 35mm film or aps size, not something like medium format or larger. Of course, there is hasselblad medium-size digital back, but it is not often seen.
  - klodolph 5y ago
    The parent comment explicitly mentioned medium-format. The point when digital became competitive with medium-format was an important milestone--it's just that if you did events, sports, wildlife, or photojournalism you were probably already on digital.
- jrockway 5y ago
  My experience mirrors yours. 50MP off of medium format and a flatbed scanner is easy. 1980s lenses look great, even with color film. I have one shot I took off the roof of a building, and you can read street signs many blocks away in the final scan. With 4x5, 100MP is pretty achievable, and even with very "inferior" lenses, the resolution is incredible. I have a picture of the Manhattan Bridge from the Brooklyn side, with the main focus being the bridge... but if you zoom in on the background, you can again read street signs. I love this kind of photo, where you can blow it up to wall size and think "nice" but walk closer and uncover tons of details that you could never see in person. It impresses me every time. Digital is in an awkward spot where semiconductor yields make larger sensors impractical, so they have to focus on hyper-optimizing lenses. This ends up being very expensive. And, there are diminishing returns; the wave nature of light introduces interesting artifacts the smaller your pixels get, so there is unfortunately a hard limit. Compare this to a big piece of plastic dipped in some chemicals, and you wonder how much we really benefited from the switch to digital. It's great for the snapshot market, it's great for people that need to produce 1000s of wall-sized photos a day, but for the large middle ground, it feels like technology for technology's sake. (I of course own a DSLR, being very into technology for technology's sake!)
  - wyager 5y ago
    > the wave nature of light introduces interesting artifacts the smaller your pixels get The particle nature as well. Because light arrival is a poisson process, your pixels need to be big enough that their “backing capacitor”, so to speak, can fit at least a few tens of thousands of electrons for acceptable noise levels. If your pixel capacitor maxes out at 100 electrons, your SNR will never exceed 10, even under optimal conditions (or, at least, you have to downscale, and then you would be better off with bigger pixels in the first place).

by _ph_ 5y ago

klodolph 5y ago

As someone who sometimes shoots medium format film, sometimes shoots digital—I am a bit skeptical about claims that digital “easily beat the resolution of classical medium format film”. I shoot with a nice, but old, MF camera and prime lens on a tripod and then do wet prints (usually) or drum scans (rarely). Based on close examination of drum scans, I estimate my setup gives images of around 60-80 megapixels. Knock that down somewhat for color. This is a bit lower than the quoted resolution for the film I use, if you read the spec sheet. Rough comparison of digital vs film leads me to conclude that you can’t just compare the camera resolution against the film scan resolution. Maybe it’s the low-pass filter / Bayer filter, but a film scan will have the same image quality as a raw digital photo of somewhat larger size (say, 50%). So if you take a high-end full-frame digital camera with a 24x36mm sensor, you get 50 megapixels, and my film setup will get you around the same number of pixels when shooting color, but the film scans will be sharper. The film is 4x the area—56x70mm. So it’s close, and I wouldn’t be surprised if a top-of-the-line full-frame digital camera could beat a 645. But 645 is the smallest of the medium format sizes, and digital wouldn’t be beating it by a huge margin. It’s a bit subjective, because the transfer functions for digital and film are so different and there’s no obvious correct way to compare them mathematically. That said, digital will easily win at higher ISOs.

leemailll 5y ago
I suspect when people compare digital against film, they often mean the 35mm film or aps size, not something like medium format or larger. Of course, there is hasselblad medium-size digital back, but it is not often seen.
- klodolph 5y ago
  The parent comment explicitly mentioned medium-format. The point when digital became competitive with medium-format was an important milestone--it's just that if you did events, sports, wildlife, or photojournalism you were probably already on digital.
jrockway 5y ago
My experience mirrors yours. 50MP off of medium format and a flatbed scanner is easy. 1980s lenses look great, even with color film. I have one shot I took off the roof of a building, and you can read street signs many blocks away in the final scan. With 4x5, 100MP is pretty achievable, and even with very "inferior" lenses, the resolution is incredible. I have a picture of the Manhattan Bridge from the Brooklyn side, with the main focus being the bridge... but if you zoom in on the background, you can again read street signs. I love this kind of photo, where you can blow it up to wall size and think "nice" but walk closer and uncover tons of details that you could never see in person. It impresses me every time. Digital is in an awkward spot where semiconductor yields make larger sensors impractical, so they have to focus on hyper-optimizing lenses. This ends up being very expensive. And, there are diminishing returns; the wave nature of light introduces interesting artifacts the smaller your pixels get, so there is unfortunately a hard limit. Compare this to a big piece of plastic dipped in some chemicals, and you wonder how much we really benefited from the switch to digital. It's great for the snapshot market, it's great for people that need to produce 1000s of wall-sized photos a day, but for the large middle ground, it feels like technology for technology's sake. (I of course own a DSLR, being very into technology for technology's sake!)
- wyager 5y ago
  > the wave nature of light introduces interesting artifacts the smaller your pixels get The particle nature as well. Because light arrival is a poisson process, your pixels need to be big enough that their “backing capacitor”, so to speak, can fit at least a few tens of thousands of electrons for acceptable noise levels. If your pixel capacitor maxes out at 100 electrons, your SNR will never exceed 10, even under optimal conditions (or, at least, you have to downscale, and then you would be better off with bigger pixels in the first place).