Scanning Color Negatives With Expanded Dynamic Range & Contrast (via DSLR & Scanner)

[JasonF]

Usually, when I scan color negatives (either with a DSLR or scanner), I end up with a scan that has a histogram that is quite compressed. What one has to typically do is adjust the levels so the histogram has better distribution or tonal range. I think they call this dynamic expansion? Anyway, this does increase the contrast, but from my understanding, may leave gaps in-between do to the expansion?

So my question is, how could I scan a color negative with an 'already' expanded tonal range (hardware wise and not necessarily via software) instead of one that would be typically compressed?

Is the reason the scanner can't capture an expanded tonal range have to do with the light source? Better light source, maybe I'd have better tonal range distribution?

Yet, maybe I'm misunderstanding something: is it that even though the tonal range is compressed, the scanner is really capturing everything that it could from the color negative? Meaning, even though in the end I must expand the tonal range, that really I'm not loosing anything in the process because all the data is really there despite what I am seeing in the histogram?

When I see from a normal scan in the histogram the tonal range distributes from say, 150 to 250 (from a histogram range of 0 to 255), it sort of bothers me. I feel like I'm missing data as it doesn't cover the full tonal range of the sensor. I'm not sure what to make of this?

[Peter De Smidt]

Don't worry about it. Shoot raw and adjust the contrast in your raw processor.

[onnect17]

Search for the specs of any c41 film and look at the densities for each color, including the gray point. Draw in paper a graph with a linear and log distribution. Once visualized then is easier to understand why some scanners include a "log" mode for negatives.

[fishbulb]

I agree with Peter - in most cases it doesn't matter. And as onnect17 said, there is a difference between linear and log mode. Unfortunately, not a lot of programs offer log histograms or curves.

Anyway, unless the scene had an extremely wide range of exposure values, the linear histogram is typically compressed in the middle. The reason is that a decent scanner, or a DSLR in RAW mode, can typically capture a wider range of EV values than most photographers like to shoot. That is, most photographers prefer scenes with a relatively low range of EV's.

For example, most of my photographs have only a 4 to 6 stop range of EV's on my spot meter. Easy for a scanner, and it definitely has a compressed histogram because the scanner can handle more range of EV's than I'm giving it.

Occasionally I will have images with a ten stop or more exposure range, typically when shooting into the sun or just with bright sun. These images are more likely to touch the edges of the histogram even during a wide gamut scan... but haven't had one blow out to solid black or white yet though, mainly because I don't like to shoot that way. Some people will scan a high-range negative twice - once for the highlights and once for the shadows - and then merge the layers later, if they are afraid that it is too contrasty for the scanner to handle in one scan.

My normal workflow for Howtek 4500 scanning is to do a preview scan with a wide gamut (using Aztek DPL), then use the curves tool to move the black and white points inward to the edges of the histogram, setting the contrast for the final scan. Then I make further adjustments to the final scanned file if needed. But not all scanning software will let you do this. If it doesn't, just scan the whole thing wide gamut, as a raw file or uncompressed TIFF, and then make edits later.

In some scanning software, you can use a custom profile for a certain type of film - for example, if you are scanning a ton of Porta 160 and you have done enough wide-gamut scans that you know where you need to set the black and white points, color curves, etc. You create a profile for the Portra 160, and scan away. But for me, I like adjusting the curves/histogram for each image separately before I do the final scan.

EDIT: If you want to play around with linear or log histograms, there is some free software for that. The page also has a good explanation of the differences: http://www.guillermoluijk.com/tutorial/histogrammar/index_en.htm

[onnect17]

Just to be clear, I was referring to the data acquisition, where some scanners hardware can change the gamma to follow a linear or log response before quantization. That's the place where the info in the emulsion gets f*. To add to the confusion, in the case of Howtek/Aztek scanners, a linear capture of densities (0=4095, 1=3072, 2=2048, ...) is called log mode.

In the case of C41, the gamma is different between color channels, so the middle point is all over the place. At the end you end up with the equivalent of light tones in the blue channel being 8 bits while the dark tones in the red channel close to 14 bits. Another issue is the resolving power in the red channel, which is much lower than the green and blue. All considered, getting an accurate shot in color negative is way more challenging (IMHO) than E-6, even with the ideal hardware.

On the other side, regarding linear and log representation, PS (and most others apps including DPL) match the center of the histogram to the gray point (or 18% reflective, or .75 density). That's is not exactly a log10 distribution but is quite close to our perception of luminance and the purpose is to match the gamma of the output device, i.e. display.