Mechanische camera's

Kijk op het fotografisch universum door Erwin Puts

What can we see in print?

I have regularly reported in the recent past that the classical AgX film emulsion surpasses digital capture in definition and resolution. Some of my readers commented that it may be nice to see more details under the microscope, but that does not count much when we look at pictures in print. It is well-known ( I have noted this myself on several occasions) that chemical-based photography has a workflow where degradation of the image is the norm and that in digital image-processing image enhancement is the norm.
It is also well-known that most pictures today are viewed on screen at 72 pixels/inch. But the computer screen is no substitute for a true print. A physical print (inkjet or chemical is still the litmus test for photographic expertise! That we do not do this in many cases is just laziness and another indication that photography is dead!

In order to analyze this proposition I set up a simple test:
Use a test chart and photograph this one with the same lens (Summilux-M asph 1.4/50mm at f/5.6) fixed to the M7 and M8 at the same distances (1, 2 and 3 meter).
The film used was Ilford Delta100 developed in Schain HRX-3 and processed on Ilfprd Multigrade IV with Heiland Splitgrade, Leitz V35 and Schneider 2.8/40mm at 16x enlargement.
Digital files were developed with RAW developer on the Mac and printed on Canon 9500 Mk2 on platinum paper.
The images with 20x, 40x and 60x negative enlargement were printed on A4 and also at maximum enlargement. This is easier with digital than with film!
The resulting prints were then scanned at 2400 resolution with the Epson V700 and transformed to web-based viewing with Graphic Converter. This last step reduces the differences that are more clearly seen in the original prints than in the web-based reproductions you see here.

First of all see here the original file at 3 meter distance.


Here you see one of the testcharts at original scale.
Note that there is a significant difference in reproduction between the black on white and the white on black figures.


Here you see the 20times reproduction of the testchart on digital (top) ) and film below). The scanning procedure has reduced the quality of the chemical print and the original is just crisper! But we may say that both prints deliver about the same quality impression.



At 60x reduction we do note a significant difference in quality: the digital print starts to lose definition and resolution (note specifically the smaller print) and the artificial enhancement blocks up the edges. Note too that the color differences are better reproduced in digital files than in panchromatic film. But the color blocks are really harshly reproduced in digital and smooth in AgX.
Top: digital; below: film



Both these pictures do not show the true differences between the film and digital capture.
See below a section of the digital file and below this the section of the scan of the print of the film negative: note in particular the shape of the printed letters.




It is clearly visible that even a 100ISO film has more reproduction capabilities than a 10 Mp sensor, specifically at major enlargements. Film holds fine detail to a larger extent than digital does and can indeed reproduce this detail in print (at least at 15+ enlargements).
Other experiments (see the Zeiss site) indicate that the differences between a 12 Mp and a 24Mp sized sensor are less significant than often assumed.
So we need at least a 40Mp sized sensor to visibly surpass the 100ISO film emulsion properly exposed and printed on paper.
When will we see the Mxx with 37.5 Mp sensor? Only then are we able to surpass the quality of the film emulsion at bigger enlargements or finer reductions of the scene.
Most current tests use testcharts that require a small distance between camera and object: this does distort the true quality differences between film and digital. This test also indicates that only a physical print can do justice to the real differences between film and digital printing. But who does his home work these days?

Update (July 14, 2009)

I got an amazing number of comments that fall in two categories.
One is that I should have used Capture One and not Raw developer, because CO is supposed to do a much better job on Leica DNG files than Raw Developer. Here we see the impact of myth! CO is the preferred developer by Leica and indeed you will find on the internet may claims that this software does the best job. I beg to differ! The software is OK, but not better than several others, including Lightroom to name one. The CO software is quite good in developing M8 files as far as colour i concerned and for BW images with subtle tone differences. The sharpness processing is just good.
See below the same image, section, but now processed by CO. The image has been directly processed and not (as the others above) gone through the print and scan process. Even with this advantage the quality is not significantly better.


The second type of comment notes that it is unfair to compare the M8 with a Bayer pattern to a BW film without one. I should have used colour film. However: colour film does not have a Bayer pattern either.
What I compared are two different workflows: one completely digital and one completely chemical, but both with the same end result: the print! And not the direct comparison on screen. Naturally you want to get the best results with either workflow, so it is valid to use BW film in one cas (AgX) and digital development and printing in the other case.