Real World Test: Kodak Gold 200 vs Nikon D60 Dynamic Range

1. What film? For example, low-contrast print film has greater DR than slide film's low DR but richer gradation in recorded tones.
   
   
2. What film format? Larger formats give greater film to image ratio so grain is less detectable at film's limits of exposure.
   
   
3. What sensor? The more convenient pocket digicams use smaller sensors which generate more sensor noise.
   
   
4. What scanner? Variations in optics, sensor resolution, scanner DR, and precision of the analog to digital conversion circuit can make a huge difference.
   
   
5. What counts as image and what is noise? This question defines the limits of DR within a single photograph, and may vary with subject matter.
   

All of these issues affect DR, so no one test can prove generally that digital or film has more of it, any more than a single test can prove cars faster than motorcycles (or even horses, in rugged terrain). In either case, the best we can do is operate specific models under circumstances as similar as possible, see which comes out ahead and by how far, and after many such trials begin to generalize differences between the technologies. Thus I set out to conduct a trial of one digital and one film camera on a day not long ago.

The Test
What follows is my admittedly imperfect attempt to compare film and digital DR and "see for myself." In comparing one representative digital and one film camera, I have tried to hold as many factors constant as practical. To that end, I printed a step wedge target (available below), taped it inside a cardboard box, and headed for a local camera shop on a cloudless midwestern summer afternoon.

Upon arrival I announced my intentions and requested a digicam of moderate cost that could store images in raw format on my Secure Digital (SD) card. Surprisingly, no compact digicam in the store would save raw files; the lowest cost option was an entry level DSLR, Nikon's D40. When I asked to go outside with the D40 (the better to test in direct sunlight), my salesman proposed to accompany me. While giving legitimacy to my activities, this also constrained my ability to experiment as the salesman insisted on "helping" me configure the camera. Worse, once we had defeated the D40's automation (ASA, aperture, auto flash, etc.) and to set it to "raw" mode, it became apparent that it was capturing only monochrome images. The salesman's struggles to defeat this having failed (and salesman plainly discouraged), we switched to the next floor model up, a Nikon D60.

Having deautomated the D60 to the best of our abilities (no simple matter; see the 200 page owner's manual if in doubt), I was finally free to photograph. Or so I thought: the sight of a grown man sitting on the sidewalk aiming a camera into a cardboard box proved irresistable to an elderly pair of woman shoppers and we were forced to pause briefly and acknowledge their incredulity.

Back to the test. I had locked the D60 at aperture F11 and film speed at ASA200, disabled the flash and set aperture priority. Now the light meter wanted to follow the autofocus and autofocus didn't want to focus on the step wedge target (too dark?)! In the name of expediency I settled for center weighted metering rather than attempt to manually meter and focus the D60.

The product of our efforts was 10 NEF files on my SD card, most of which were out of focus or badly exposed, but a few acceptable for this test. I thanked the salesman for his help, and found the next part of the experiment much easier: set film camera aperture and shutter, aim, focus, shoot, advance film and repeat. I ran the machine rather than it running me and finished this part of the test in no time. The experience of using these two cameras could not have felt more different.

The Results
Below are two images of the same scene outside the camera shop, taken within a few minutes of each other. The subject is a step wedge target taped inside a cardboard box with a backdrop of the shopping center in full sunlight. The first image was obtained with the Nikon D60, its 18-55mm VR lens set as near as I could approximate to focal length 40mm, saved to this raw (NEF) file. Settings recorded in the file are:

Make: "NIKON CORPORATION"
Model: "NIKON D60"
Exposure (sec): 0.0167 (about 1/60 sec)
Aperture (f): 11.0000
Focal Length (mm): 26.0000
Exposure Bias (stops): 1.0000
Metering Mode: 2
Exposure Program: 3
Flash: 0
Film Speed (ISO): 200
Light Source: 10
Sensing Method: 2
Custom Rendered: 0
Exposure Mode: 0
White Balance: 1
Digital Zoom: 1.0000
Equiv 35mm FL: 39
Scene Capture: 0
Gain: 0
Contrast: 0
Saturation: 0
Sharpness: 0
Subj Dist Range: 0
Software: "Ver.1.00 "
File Source: "03"
Scene Type: "01"
CFA Pattern: "0002000201020001"

The D60 has the following characteristics:
Effective pixels: 10.2 million
Image sensor: 23.6 × 15.8 mm CCD sensor
File Format: NEF (RAW): 12-bit compressed

The second image was obtained using 35mm Kodak Gold 200 print film with a Leica 40mm FL Summicron C lens, having the film processed at Walgreens, and then scanning the film with a Nikon Coolscan IV driven by VueScan scanning software. By coincidence, the Coolscan IV and D60's NEF both store data in 12 bits of precision.

Both cameras were set to ASA 200 with an F stop of 11, median exposure was as metered by each camera. I bracketed several images with both cameras, and performed all post-processing with VueScan, which can process .NEF files in the same way as scanner data. Using VueScan I adjusted each image so the step wedges were as nearly identical as possible. To accomplish this without loss of information I first set white point to .002 for each image, then used black point, brightness, and white balance to produce visibly similar output with minimal pixel clipping according to VueScan's linear color histogram (below). Next I chose the pair of images in which the values in the step wedge were closest.

In summary, my results were as follows:

	Nikon D60	Kodak Gold 200
ASA	200	200
Aperture	F11	F11
Lens	Nikon 18-55mm VR	Leitz Summicron-C 40mm
Focal length	39mm (equiv. 35mm)	40mm
Hardware color precision	12 bits	12 bits
Imaging area	23.6 × 15.8 mm	36 x 24 mm
Digital conversion	Direct	From film with Nikon CoolScan IV
Image processing software	VueScan Pro	VueScan Pro
VueScan sharpening	None	None
VueScan white point / black point / brightness controls	.002 / 0 / 2.08	.002 / 1 / 1.18
VueScan linear color histogram

Since I matched on the step wedge, the significant differences below are in what appears outside the shadow box. Both images are overexposed in this area, but the sidewalk and sky areas in the D60 image are more washed out than in the Kodak Gold 200 image:





Below are full resolution crops of the above two images in the shadow area, to display film grain versus sensor noise in this, another key area related to DR:





Analysis
While darker areas of the step wedge show roughly equal noise, lighter areas of the D60 image are cleaner. However, these lighter areas also exhibit less contrast than Kodak Gold, squares 25 through 45 being nearly equal to their neighbors on the right. So while we have less noise, we also have less contrast information in the D60 image, hence less effective range. Either image can be post processed to increase contrast, but this increases noise also.

In this test a high-end consumer digicam falls short of normal consumer color print film in the area of DR. Granted I didn't let the D60's ASA "float" or use Nikon's "Active D-Lighting" in this test; either of these may have avoided highlight blowout. But I also did not use the lowest contrast film available, professional film processing, or the best available scanner. So I feel that while far from perfect, the test is not unfair.

Does this "prove" film has greater DR than digital? No, but it may indicate consumer print film has a DR advantage over a relatively large, expensive, and complicated digicam. I think it tends to disprove the idea that digital has an obvious DR advantage over film, as I expect a comparison with a typical (smaller sensor) consumer digicam would show even greater DR superiority for film. Would a larger digicam's DR have exceeded Kodak Gold 200? Quite possibly, but so would professional grade film, or Kodak Gold in a larger format.

Followup
For me the upshot of this test is, the DR question is far from closed. More testing is needed, which is where YOU come in: don't rely on someone else's findings: get out there, test, see for yourself, and please report back. As I've shown, you don't even need to own the equipment!

To this end, here is the step wedge image file I used to generate my target. I printed it in color with a Canon ip8500 inkjet printer on HP high resolution paper, setting the printer preferences to "high resolution paper" (high resolution or better paper is needed to produce solid black tones). I'll publish any thorough results submitted here; please include:

• Type of digital and film equipment used (camera, lens, film)
  
• Settings used (ASA, focal length, aperture, shutter speed)
  
• Environmental information (lighting temperature, intensity, and consistency)

Resources
For more on Dynamic Range, see Dante Stella, More is Less Is More: The Devil is in the Dynamic Range.

For insight about measurement of density range and bit depth, I also recommend Wayne Fulton.

For a filmmaker's perspective on the DR controversy, see Juan P. Perierra, Dynamic Range, Latitude and the Quest for Digital Film.

R. N. Clark, Dynamic Range and Transfer Functions of Digital Images and Comparison to Film is one of the most frequently cited web resources on the topic of film versus digital DR. Comparing a professional digital camera with 35mm film, Clark came to the conclusion that "Digital cameras, like the Canon 1D Mark II, show a huge dynamic range compared to either print or slide film, at least for the films compared."

Ken Rockwell, Film vs. Digital Cameras is also frequently cited. Rockwell comes to a different conclusion: "CCDs and the related capture electronics will need about ten times more dynamic range (three stops) than they have today to be able to simulate film's shoulder....This is the biggest image defect in digital cameras today."

---

That's all for now. To all, thanks for reading and happy image acquisition, by whatever means! - Carson Wilson