Looking for link to Epson 4990 scanned film to one shot with Nikon D100

[sanking]

All of the studies noted are interesting, though it is impossible to draw hard conclusions since they use different methodologies and each appears to have an agenda.

There is no question in my mind but that 12mp DSLR easily beats 35mm film when the scan is made with a prosumer scanner. With a drum scan and the right film, 35mm might hold up. And there is no question but that 12mp - 16mp DSLR is capable of outstanding professional results, and that there are many valid reasons why professionals prefer the DSLR work flow to scanning film.

However, if the question is whether 12mp DSLR beats 6X7 and 6X9 formats when the film is scanned with a prosumer scanner like the Epsons 4990 or V700, my experience is that the DSLR has an advantage in grain when compared at the same ASA, but film appears to win in detail.

Attached are two small jpeg images. They represent a small crop from a 12mp DSLR and a scan of a 6X7 film shot on Mamiya 7 with 65mm lens, both shot on a tripod. The field of view with the DSLR was adjusted to match that of the Mamiya 7 at the shooting location. The center of the image, which is about 16 feet away from the camera, contained a resolution target. The film was Fuji Across developed in Pyrocat-MC 1:1:100.

If you examine the film with a microscope it is possible to separate the resolution bars down to Group 0 - Element 3, which indicates resolution of about 85 lppm on the film. I am able to capture about 60 lppm of the potential 85 lppm with the EverSmart Pro scanner, but even this high end scanner still leaves on the table 25 lppm of resolution. To pull all the detail out of this negative would require a very high end scanner capable of at least *effective* 4500 ppi.

How does the 12mp DSLR compare to the scan of the film with the Epson 4990? Well, neither come anywhere close to the actual detail on the film, and nowhere close to the scan made with the EverSmart Pro.

As for the rest, judge for yourself from the attached images. Bear in mind that the crop represents a very small part of the original frame, and in practice it would be possible to make very nice looking 13X19" prints from either the DSLR or the prosumer scan of the 6X7cm negative.

Sandy

[jwaddison]

I know zip about digital, but didn't I read about an article originating from Kodak that basically said there is much more information on film than on digital files originating from cameras or scans? If I remember correctly I think the piece I read quoted the Kodak scientists as saying that there were 100 megabytes of info on 35mm movie film - which apparently would translate into 200 mb on still camera 35mm film. If I'm remembering correctly, and if Kodak was quoted correctly, then the comparisons above are only valid for digital cameras vs scanners, not digital cameras vs film.

[walter23]

His more recent D200 comparison is kind of cute:
http://www.kenrockwell.com/nikon/d200/d200-vs-4x5.htm

There's truth to what he says about having the opportunity to get shots that he wouldn't get with 4x5 - but I kind of enjoy that about 4x5 now. I don't have to try to capture every possible shot, and as a consequence I value my fewer shots more (and I have less crap to sort through on memory cards when I get home).

[sanking]

I know zip about digital, but didn't I read about an article originating from Kodak that basically said there is much more information on film than on digital files originating from cameras or scans? If I remember correctly I think the piece I read quoted the Kodak scientists as saying that there were 100 megabytes of info on 35mm movie film - which apparently would translate into 200 mb on still camera 35mm film. If I'm remembering correctly, and if Kodak was quoted correctly, then the comparisons above are only valid for digital cameras vs scanners, not digital cameras vs film.

Well, yes the comparisons are only valid for the specific digital camera and the specific scan with an Epson 4990. I did mention that this camera and lens (Mamiya 7 with 65mm lens) produced resolution of about 85 lppm. The Epson scanner does not come anywhere close to pulling all that detail out of the negative. For that it would take a drum scanner working at about 5000 ppi.

Just to show how superior a high end scanner is to the Epson I scanned the negative again with my LeafScan 45 at 5080 ppi. As you can see, this scan is vastly superior in terms of detail to both the 12 mp digital image and to the Epson scan, though it does not quite resolve all of the detail on the negative, which shows separation of the bars in Group 0 - Element 3.

The Achilles heel of MF is the lack of inexpensive high quality scanners. MF is potentially far superior to even 20 mp DSLR, but in order to get that kind of quality you will need a very expensive scanner. However, my point is that there is not that much difference between 12mp DSLR and MF even when the scan is done with a prosumer scanner like the Epson 4990.

Sandy King

[Marko]

Sandy,

I am submitting the following as a question, not a statement!

If I understand the implication of your comparison, it is not just the quality that film is capable of delivering, it is primarily the question of the entire chain of factors starting with capture and ending with the print - or whatever final form of the photograph one chooses, for that matter.

If the end product is a print, I see three possible (simplified) paths to it:

1. Film capture - traditional enlargement.
2. Film capture - scanning - computer print (any of the available options).
3. Digital capture - computer print.

The first option would include two optical paths, each introducing its own set of distortions and degradations, having a compound effect in the end. How much of the original 100 lppm that film is theoretically capable of would remain at the end? Mamya 7, which is arguably one of the finest possible tools within its format delivered 85 lppm. What about the enlarging lens? What are the best enlarger and lens capable of delivering out of those 85 lppm?

The second option replaces the enlarger/lens combo with the scanner and its own set of limitations and distortions.

The third option replaces the limitations (or capabilities) of film with those of a sensor. Assuming that the lens used is comparable in performance to that of Mamiya 7 (high standard but not unreachable), apparent inferiority of the sensor as compared to film would be mitigated to a very measurable extent by the fact that there are no other optical distortions in this path. Another advantage of this option is the absence of noise.

The final set of limitations that all three print options are faced with is the resolving ability of paper (or printer/ink/paper combination in the last two options).

It seems that comparing entire paths would make more practical sense than comparing isolated elements. And even then, I can see at least two, possibly three distinct, possibly even concurrent aspects of such a comparison: quality, economy and workflow. Did anybody perform such a comparison?

[Kirk Gittings]

my point is that there is not that much difference between 12mp DSLR and MF even when the scan is done with a prosumer scanner like the Epson 4990.

I agree completely with Sandy's conclusions and would like to add that the situation does not change when using the next generation of prosumer flatbeds like the Epson 750.

Not that it would change the outcome, I am curious about which lens you were using on the 12MP test.

[sanking]

It seems that comparing entire paths would make more practical sense than comparing isolated elements. And even then, I can see at least two, possibly three distinct, possibly even concurrent aspects of such a comparison: quality, economy and workflow. Did anybody perform such a comparison?

Marko,

To some extent the comparison studies done at http://www.diax.nl/pages/start_mamiya_nikon_uk.html and at http://www.luminous-landscape.com/re...shootout.shtml do more or less what you suggest. Unfortunately they reach rather different conclusions.

Rather than look at the entire path I chose to consider only the absolute potential of the two systems to capture detail. How much detail is required is somewhat subjective in that it is subject dependent and it must also be considered in relationship to final print size.

Bottom line is that if you print no larger than 13X19” with an inkjet printer a 12mp DSLR will give you all the quality you need.

If you print larger, or crop, or print on a Lambda, a high quality 6X7 or 6X9 film camera, with a drum scan, clearly has the advantage.

Sandy King

[sanking]

Not that it would change the outcome, I am curious about which lens you were using on the 12MP test.

Kirk,

I used a Canon 5D with 24-105mm zoom. Would have used a fixed focal length lens had one been available, but this was a loan camera and the zoom was the only one I could get.

Sandy

[Ben R]

Kirk will back me up on this as it's a position that he has held for a while but to maximise the resolution of the 5D you really need the very best in glass. My 24-70L is an exceptionally sharp and contrasty work horse but the 85L stopped down a bit makes it look like a cheap milk bottle in comparison. I've never used Leica and Contax primes on it but I can only imagine how limited my ability to understand the real resolving power of the camera is due to my lack of use of the very best glass.

I had a 4990 and to be honest even when trying different heights for the film I couldn't get more resolution out of a 4X5 neg than high res imacon scans from my 645 negs and trannies. There certainly wasn't a huge difference between the 5D and those scans.

Because of that I came swiftly to the conclusion that it is useless to make comparisons based on consumer equipment or even some pro level equipment.

So why do we do it?

Because testing has a huge level of subjectivity built in because we are doing it relative to our workflow. Yes you might beat a 5D with a drum scanned 35mm neg and certainly with a drum scanned MF neg, but I can get that quality every time I press the shutter at 3FPS if I so choose and it doesn't cost me a penny extra. That makes a significant difference to our perceptions of a true comparison. If I spend a lot of money and have a top drum scan then I can beat my DSLR, but I don't have the money, the time, the patience, the hard drive space, etc so to me that is irrelevant, I want to compare on my level not a theoretical level alluded to by purists.

Reichmann said that the 1Ds outresolved 6X7 but when he did his big comparison he admitted that the 645 scan outresolved the 1Ds mkII. When I asked him about the disparity he said that his original comparison took into consideration the whole image and that he said the 1Ds image as a whole beat the 6X7 (noise, etc). Needless to say that wasn't true, he specifically compared resolved detail and wouldn't admit what was written in black and white. Any respect for his opinions that I had had dissolved at that point. I had an original 1Ds and have two 5D's and I think that my 5D with exotic primes can match an imacon scanned 645 but not more than that. The 1Ds couldn't come close.

That said, when I wanted big print sizes I chose LF not MF. If I was going to move away from the ease, convenience and resolving power of the 5D then I wanted a jump big enough to be worth the hassle and cost. 4X5 was it. I am however going to be sticking with drum scanning, the 4990 was like having a huge pile of candy and only being able to eat a tiny bit of it!

[Ron Marshall]

I agree with what Sandy and Kirk and Ben have said. I get wonderful detail from my Mamiya 7; however, most of the images I scan on a 4990 and print only as large as 11x14. For any larger size I get a drum scan.