semilog
curmudgeonly optimist
Here's a 13.5 megapixel scan (4500 x 3000) of, if I recall correctly, an ACROS negative. The camera is an M6 and the lens is the ZM 35/2.8. Probably f/11. The resolution of this image is mainly limited by the scanner used. In addition, most of the "grain" visible here is actually due to diffraction off of the actual grain clumps — which are visible in a microscope but too small for this consumer scanner to resolve.
This is a handheld, not tripod, shot. Certainly not focused with maximal care. Atrocious backlighting. Probably overexposed. Scanned on a relatively cheap scanner. Sloppy. I wasn't even really trying for maximal resolution. Even so, the resolution (the ability to separate line--pairs) is at the limit of the pixellation, at greater than 12 megapixel resolution.
And again, I emphasize that the limitation here is the scanner, not the film or the lens. Here we have a sampling grid of 125 pixels/mm. That translates a maximum possible scan resolution of 62 cyc/mm. Which is what we're actually seeing — or darned close to it. An enlarger print with a good lens (an apo-Rodagon, say) should readily resolve at least twice that. Meaning that a really good monochrome scan of this negative would require on the order of 50 megapixels (long edge = 9000 pixels).
In summary: 100 lp/mm on 35mm film (about 38 megapixels) is not heroic. It should be achievable under reasonable real-world conditions with 2TMY. With TMX or ACROS, a reasonable lens, and good technique, 100 lp/mm should be routine.
You were saying…?
Scrambler
Well-known
Knowing now what can happen you might upload. Or the internet connection from Uganda might still be unreliable and slow (it is in the only developing country I have experience with, Papua New Guinea) so perhaps you wouldn't use the cloud but would have made duplicate electronic copies.
With extra caution, even with film you would probably now separate your collections, use sealed metal containers and carry at least some in your personal luggage. Perhaps mail to a friend in smaller parcels over time?
We have all had electronic media crash but the risk this poses can be reduced with redundant backups. I agree that in some ways physical media are more fragile: backups lack the quality of the original and everything is subject to the risk of fire, theft, insects etc.
Ansel
Well-known
Even if digital resolution did beat that of film you still end up with a very high resolution artificial image. The virtue of film is that it is literally a physical reflection of the origina scene, not a recreation by a computer chip. The digital images above just look artificial.
semilog
curmudgeonly optimist
Here's another example. The scan here was 21 megapixels (5500 on the long edge), but the result is somewhat lower-resolution. The pixels on this scan are 6.4 µm center-to-center (155 px/mm).
In this picture we get something like 40 lp/mm. The film? Good old Neopan 1600.
Tri-X will do at least as well. Rochester says the MTF-50 is 50 cyc/mm and the peak resolution is a bit higher, maybe 70 lp/mm. And the pictures I've shown in this thread demonstrate that resolutions close to published numbers are readily attainable by any schmuck with a decent lens and some D-76 or XTOL. They are not, as some on this thread have suggested, merely airy-fairy laboratory values.
There is a reason why film and lens and sensor companies do science: it works.
So a Monochrom can beat the resolution of Neopan 1600, but not of ACROS. And its resolution is roughly the same as expertly-handled Tri-X.
Of course, the Monochrom has other virtues. It can shoot creamy, nearly grainless pictures with good DR at up to ISO 3200…
In this picture we get something like 40 lp/mm. The film? Good old Neopan 1600.
Tri-X will do at least as well. Rochester says the MTF-50 is 50 cyc/mm and the peak resolution is a bit higher, maybe 70 lp/mm. And the pictures I've shown in this thread demonstrate that resolutions close to published numbers are readily attainable by any schmuck with a decent lens and some D-76 or XTOL. They are not, as some on this thread have suggested, merely airy-fairy laboratory values.
There is a reason why film and lens and sensor companies do science: it works.
So a Monochrom can beat the resolution of Neopan 1600, but not of ACROS. And its resolution is roughly the same as expertly-handled Tri-X.
Of course, the Monochrom has other virtues. It can shoot creamy, nearly grainless pictures with good DR at up to ISO 3200…
Highway 61
Revisited
Come on... Look here (scroll down the page to see the D700 file).Then, just imagine what a very up-to-date 16MP, 24MP or 36MP 24x36 sensor would offer in front of a 4000dpi scan of a FP4+ 35mm image.
There's no shame for film lovers at acknowledging that digital outperforms film in that matter.
Corran
Well-known
Classic fallacy.
There isn't anything close to the resolution you claim in those images. And you can't claim to say it has XXlp/mm with no actual measurement.
Your scanner DPI isn't a measure of resolution. Over and over I hear this - "oh my Epson scans at 6400 DPI so 35mm film has 40mp of resolution" or something like that. It's just wrong.
Take any of those again and a comparison shot with a D700 or any other similar/newer DSLR and compare.
semilog
curmudgeonly optimist
Then, just imagine what a very up-to-date 16MP, 24MP or 36MP 24x36 sensor would offer in front of a 4000dpi scan of a FP4+ 35mm image.
There's no shame for film lovers at acknowledging that digital outperforms film in that matter.
Once again: a (nominally) 4000 dpi consumer scanner (it will really be a lot worse) cannot retrieve all the detail in the film. It is hobbled by film flatness and by the scanner optics. The "grain" you're seeing in the scans is not the actual film grain. So you're tying one of the negative's hands behind its back, and then punching it.
The MTF of the D700 is indeed higher at the sensor's ultimate resolution meaning that it will be better at preserving fine, low-contrast texture. But a good film (especially TMX or ACROS), printed optically, is capable of higher ultimate resolution.
And the D700 cannot get to 4000 dpi at all. It maxes out at about 2800 dpi. Which is why the D700 image is smaller than the others in the comparison you've linked.
I'm in no way arguing that film is better than digital. It's not. But once again: film, handled carefully, is considerably better than people tend to remember.
Indeed, the comparison you've linked hints at the real problem with film: the length of the imaging chain. With digital you fire the shutter, the photoelectrons do their dance to the amplifiers and the resulting voltage is converted to a number by the DAC. Bang. Your'e done. Maybe (though not with the Monochrom) you have to run the data through a de-mosaicing kernel.
With film you finish the roll, wind the film back into the cassette (hope it's not scratched), dunk it in chemistry (that may or may not be fresh and that varies with the quality of the water used to mix it), agitate it (more or less the same way as last time?), dry it (dust! More scratches!) project it through an optical system (whole new set of aberrations!) that must hold the film flat (glass carrier! more dust! newton rings!), must be accurately aligned and focused, onto a piece of paper that must be flat and then itself must be fresh and developed in fresh chemistry…
There are so many places for a film image to get messed up. The degradation is additive (in some cases, multiplicative). If you use a consumer scanner, all bets are off. And THAT is, at root, why film is difficult and digital is easy. The typical product that the typical user gets under typical working conditions is just a lot, lot, lot less degraded.
Worth noting is that film was finally and conclusively replaced by digital sensors in the field of electron microscopy only in the last 24 months. A new generation of 16 and 64 megapixel (4k and 8k) CMOS detectors has in one shot obsoleted both film and CCD sensors in that field. The CMOS detectors are (finally!) better in resolution (due to high MTF and some really clever engineering tricks that allow sub-pixel resolution to be achieved), and they are vastly better in sensitivity. The best of these detectors also cost almost $2m a piece. Our Department just got one (thanks, super-awesome private foundation!). We've finally ripped out the darkroom and its beautiful nitrogen-purge dip-and-dunk system, in 2014.
But again: the fact that it held on until about 2012 is a testament to just how good film as a detection technology was.
semilog
curmudgeonly optimist
Absolutely correct. A key point. Can't be re-stated too often.
I was reasonably careful to say that this was the nominal maximum possible resolution of the scanner, not the actual resolution achieved. The numbers there refer to the file size not to the optical resolution of the scanner.
Consumer-grade "4000 dpi" scanners pretty much never have good modulation transfer curves above 1500 dpi or so. Most of them crap out at about 2400 dpi. A few have gone to ~3000 dpi with reasonable contrast and relatively low noise, at least at moderate Dmax.
Which is why a comparison of a DSLR to film scanned on a consumer scanner will always make film look worse than it actually is.
My point is that, even with that garbage scanning step in the imaging chain, film is capable of remarkably high resolution.
semilog
curmudgeonly optimist
There are two classical resolution criteria:
Abbe's criterion, which is the smallest distance between two lines or point sources that can be resolved, and the diameter or width of the finest structures that can be resolved (MWHM = mean width at half maximum).
In the picture I linked, the finest structures resolved are ~3 pixels apart with good contrast. Thus, we're at the limit of the scanner's sampling grid. It's a crude method of estimation, but it's plenty accurate to establish the point, which is that 70-100 cyc/mm is not difficult to obtain on a negative or transparency with good film and a good lens. Getting that resolution OFF of the film is the hard part.
I do quantitative work with optical microscopy systems on a pretty routine basis, by the way. The positional measurements in the first project linked above were done with a precision of less than 10 nm. The diffraction limit of the microscope was about 200 nm. My estimates in this thread are crude, but they are not wrong. And when it's actually called for, I do know how to make a really rigorous high-resolution measurement.
Corran
Well-known
If using that criterion then the D700 images I have shot have "more" than 12mp of data. That isn't really the right way to say it but whatever.
There really isn't a point to this discussion when we only have one half of the equation (your photo) and no point of comparison with the digital equivalent. I've done that myself and it's conclusive, to me, that 35mm film just doesn't hold that much resolution, using the criterion of direct comparison to digital files.
And again, it doesn't matter to me either, with regard to using the medium. I chose to use film for other reasons (or larger film if I want the resolution).
There really isn't a point to this discussion when we only have one half of the equation (your photo) and no point of comparison with the digital equivalent. I've done that myself and it's conclusive, to me, that 35mm film just doesn't hold that much resolution, using the criterion of direct comparison to digital files.
And again, it doesn't matter to me either, with regard to using the medium. I chose to use film for other reasons (or larger film if I want the resolution).
semilog
curmudgeonly optimist
The data quality from a D700 is very high. That is, there is very little pixel-to-pixel noise. Film, especially when you're looking at low-contrast fine structures, has a more noise (more spurious information). That's why the D700 images look (are) smoother.
Though, again, the comparisons in this thread are not fair. An optical print from a good negative contains a LOT more information than what one can achieve using a consumer-grade scan. More resolution, and more density information in the highlights.
Will a D700 image look better in a big blowup than, say, a TMX negative or a 35mm Velvia slide? Maybe, maybe not. Depends on the subject, the lighting, the post-processing. The Velvia slide or TMX negative will in fact contain more fine detail, but at lower contrast and with more noise.
Here's a comparison of the data extractable from a Velvia 50 slide by two consumer scanners versus what we see in the transparency with a microscope (or a really good enlarger lens like the Apo-Rodagon). Velvia is capable of resolving more than 150 lp/mm. This is why it is never really fair to use a cheap scanner to evaluate how well film performs. Note that with the microscope, we can finally begin to resolve the individual dye clouds (on a TMX negative we would at that magnification finally be resolving the grain).
Will a carefully-captured D800 or A7R image made with a really hot lens contain more useful detail than either one? Of this, there is little doubt.
But in the sensu stricto sense used by folks working in technical photography, optics, microscopy, etc., the resolution of modern film emulsions (definitely: TMX, ACROS, Ektar 100, Velvia; possibly: 2TMY and Provia) is higher than that of any 12 megapixel sensor. That was my original claim, and it is not controversial.
Corran
Well-known
I just don't agree.
Optical prints are beautiful but it's trivial to make images with more fine detail from digital prints, with proper technique of course.
I'm sure you know way more about microscopy and everything else in a laboratory environment but I don't look or print film with a microscope.
Optical prints are beautiful but it's trivial to make images with more fine detail from digital prints, with proper technique of course.
I'm sure you know way more about microscopy and everything else in a laboratory environment but I don't look or print film with a microscope.
semilog
curmudgeonly optimist
So you haven't actually read what I wrote, then, but are instead replying to something that you imagine I might have said.
I do agree that by most quality metrics, a decent D700 capture exceeds the quality of almost all 35mm negatives. In fact, that's even true of files from my old 8 megapixel Olympus D-500. But it's not because of the maximum attainable resolution. In fact, by most reasonable image quality metrics, an A7R with the native 35mm/2.8 lens, cropped square (throwing away ⅓ of the image), will do better than a Hasselblad with a Zeiss 80mm mounted.
One more time: film can be better -- a lot better -- than most people realize: Film versus Leica M9.
But it is very hard to extract that level of performance due to the complexity of the imaging chain. Digital has a short and extremely high-fidelity imaging chain.
Digital long (several years) ago eclipsed film in signal-to-noise and sensitivity, and now has matched and in some cases exceeded film in absolute attainable resolution.
Corran
Well-known
Of course I read it.
And basically we are agreeing in some aspects.
But not all.
And basically we are agreeing in some aspects.
But not all.
sevad
Member
The results from the MM look better from a technical point of view, except for highlight detail, but I much prefer the Tri X photographs. If you want really sharp and really low noise then it's the MM. If you want to make art, it's the M6.
__--
Well-known
sevad, you better tell it to Ralph Gobson. [Just to say that this sort of pontification, in my view, makes no sense at all].
Paris
CAPTION: "i hope that's an M6 that he shot me with. I don't want to come out technically perfect."
MITCH ALLAND/Potomac, MD
Download links for book project pdf files
Chiang Tung Days
Tristes Tropiques
Bangkok Hysteria
Paris au rythme de Basquiat and Other Poems
ParisCAPTION: "i hope that's an M6 that he shot me with. I don't want to come out technically perfect."
MITCH ALLAND/Potomac, MD
Download links for book project pdf files
Chiang Tung Days
Tristes Tropiques
Bangkok Hysteria
Paris au rythme de Basquiat and Other Poems
sevad
Member
It's just an opinion and you can take it or leave it - makes no difference to me. Here's another opinion - if I see another pub snap I think I'll slash my wrists! Have you people no imagination? :bang:
noisycheese
Normal(ish) Human
+1.
The Monochrom is capable of making some very nice prints, if the printer is very adept at processing the files. However, they still do not measure up to the results that Tri-X and wet darkroom printing are able to produce IMO.
My approach has evolved to this: I shoot Tri-X in my M4-P when I want B&W prints and use my M240 for color work.
dmc
Bessa Driver
Absolute total rubbish! Dude, stick to what you know. The fact you don't own an M says it all.Even with a 4000 dpi dedicated film scanner on the Tri-X side, the difference in quality would have been much more pronounced in favor of the MM, had he not posted postal stamps-sized images, but full res. ones, and had he post-processed the MM RAW files properly, not just so that the MM photos looked like the Tri-X ones. Come on guys. I love Tri-X and still shoot it as my main film, and I don't own any MM, but the MM sensor outperforms any B&W film available now, in terms of resolution and greyscale depth. What it should be compared with is a 4x5" low speed film sheet, not a 24x36 ISO 400 one. The goal of this "test" was to demonstrate that Tri-X was "nicer", so it was biased from the beginning. Come on... naked body parts yes, but nothing nasty. More : great and nice shots. He's been published in some papers and is a good photographer knowing what he's doing. We're in 2014 - I guess your kids will see way worse before you can be aware of it...
PS - And yes, I do own an M. ANY film out resolves it, period.
sevad
Member
Film out-resolves digital only if you use the highest resolving films such as Adox CMS. The rest of the time digital will produce sharper looking prints than film and probably more resolution. If you use Adox CMS in a Hasselblad, for instance, you'll blow away digital from a resolution point of view.