I enjoyed Rich's theoretical treatment and the discussion about film vs digital!
Every theory needs to be tested for validity against experimental data. So let's do this! I took the same picture of a park in plenty of light with a digital and analogue camera, using the exact same lens and a heavy Gitzo tripod.
- Nikon D600 (full-frame 24 Mpixel sensor) + Nikon AI-S 50mm f/1.4, stopped down to f/8.
- Nikon FE2 with Ilford FP4 (ISO 100) + Nikon AI-S 50mm f/1.4, stopped down to f/8.
I then converted the 24 Mpix digital image to monochrome for comparison with the FP4 b/w film. Otherwise, the picture is completely untouched and I made no changes in Photoshop. The FP4 negative was scanned with a Plustek 8100 film scanner at 5,000 dpi. The result from the scan is a 66 Mbytes 16-bit TIF vector file which is equivalent to a 33 Mpixel JPG file (6888x4788 pixels, 8 bit). I'm therefore oversampling any information that is in the negative. Below are low-res versions of the two images (no cropping).
Next, I decreased the resolution of the original full-frame 24 Mpixel digital file to 16 Mpixel, 12 Mpixel, and 8 Mpixel so we can compare them to the scanned full-res FP4 analogue picture. I then picked a region at the column of the statue in the park (marked by a red box) where you can see a street sign in the background. The street sign says: "Capitol St NE 1100." Below are crops from these files so we can compare them. The upper row gives crops of the full-frame 24 Mpix, 16 Mpix, 12 Mpix and 8 Mpix digital images. In the lower row you see the same crop of the full-res scanned FP4 analogue image 4 times for ease of comparison with the digital images above. Let's have a look what we see in the crops:
- 24 Mpixel digital image: the street sign is super sharp; no pixels are visible even at this high zoom level.
- 16 Mpixel digital image: the street sign is still sharp; you start to see pixels.
- 12 Mpixel digital image: the street sign is still legible; pixels become prominent.
- 8 Mpixel digital image: unable to read the street sign, heavy pixelation.
- FP4 analogue image: you can barely read the street sign due to the low resolving power of the film. In terms of the amount of information that is stored in the analogue film, it is similar to the 8 Mpix digital file (you can barely read "Capitol 1100", but not "St NE"). However, in terms of overall appearance, the analogue film is somewhere between the 12 Mpix and the 16 Mpix digital image. Now, this is hard to quantify because our eyes and the brain that interprets the information are analog "image processing" organs and favor analogue information (such as film) over digital. I'd argue that you could enlarge a film image to a larger size than a digital image that has stored the same amount of information as the analogue film image because our brain prefers a bit of blurriness over pixelation.
Caveats: you might criticize that I used a film scanner (Plustek 8100) that is not on par with professional drum scanners. However, I have had negatives scanned on a drum scanner by Northcoast in "enhanced scan mode" (3339x5036 pixels) and believe that my scans are of higher quality. But to be fair, if you'd find the perfect film/developer combination to reduce grain as much as you can, and used a high-quality drum scanner, you might be able to push the results a bit further than I did.
My conclusion from this experiment is:
A 35mm ISO 100 analogue film image stores the same amount of information as an 8 Mpix digital image.
In terms of overall appearance and quality, a 35mm ISO 100 analogue film image is equivalent to a 16 Mpix digital image.
Coming back to Rich's analysis, you did a great job and my experiment shows that you overestimated the quality of film by only 25% (20 Mpix vs 16 Mpix).
Looking forward to your comments!
(click on the above image to enlarge)
