I print digital negatives from Photoshop. A digital negative is a grayscale image printed on clear film that is used for contact printing in the darkroom. The document profile is Adobe RGB, but the components R, G, and B are all equal.
I allow the printer (an Epson P700) to manage color. When I choose a Media Type in the print dialog, I choose one of the photo papers, since there is no such media type as "clear film." Since Photoshop thinks it is printing on paper, it lays down ink in such a way that the print, will match the image on my screen when viewed by a human observer in reflected light, which is to say it applies some kind of gamma correction. As a result, print density increase rapidly as tone approach RGB=0 (K-100%). In the following plot, the X axis is digital grayscale tone and the Y axis is log density.
For digital negatives, these rapidly increasing densities near RGB=0 are a problem. A more linear curve would be desirable. Is there is anything I can do to make the curve more linear? A gamma correction of 1.0 would be ideal, since it would produce a negative with linear tones. Or perhaps the solution is to let Photosop manage color and use some kind of output ICC profile that would straighten the curve. If so, can you suggest how to either find or make such a profile?
Explore related tutorials & articles
12
Replies
The problem is how your printer lays ink. It should be possible to make an ICC profile, but it is hard for you to control the density when it is the printer that is managing the colours. Ther Printer is not printing in RGB, but converting from RGB to the printers inks, this process is controlled by the printer.
I am also a little confused since you say you are doing greyscale and then you talk about RGB. You are not using true greyscale? You can set output device curves in photoshop, but not if the device is managing colours. You can make both custom dotgain and custom gama curves in your colour settings.
There is no Media type or ICC profile for clear film, and I have yet to hear of anyone who has made one, at least not for pedestrian photographic applications. I'm happy with the printer thinking it's printing on some kind of paper surface. I simply want less of a "hockey stick" effect in the deep shadows (something like below). I don't need to control the output any more than that.
The "printer" is a component of a model of tone reproduction that I'm using to simulate alternative process printing. The printer's job is to transform digital tones to transmission densities. Transmission densities control exposure. The model works mathematically as long as transmission densities increase monotonically from K=0% to K=100%, so any continuously upward-sloping curve will work, at least in theory. However in practice, if you consider error propagation, portions of the curve with very steep or very shallow slope do not sort out the corresponding tones very well. A linear curve is ideal for error propagation.
A solution found in other digital negative calibration methods is to add color to the negative using an adjustment layer or LUT (e.g., a Color Fill layer in Overlay mode). Here is an example comparing a grayscale negative with one that has been colorized with a certain shade of red:
As you can see, colorizing has exactly the effect on the negative that I'm looking for. However I cannot use such adjustments to the photoshop image because my model assumes a single gray channel, and these adjustments involve multiple channels. I have experimented with applying tints in the printer driver, but they have minimal effect compared with a color adjustment layer.
My system will be used by people with different printers who are also using a variety of clear film media. An ICC profile is specific to the printer and the media, so everyone using the system would have to either make their own ICC profile or to pay someone else to make it for them. That would not be practical.
I mentioned earlier in this discussion that I had experimented with tints in the printer driver, but had found that tints did not make a significant difference. Reviewing those experiments, I see that they were done on an Epson P400 which never developed high enough density to exhibit the dreaded "hockey stick" pattern (the p400 printer driver does not allow you to control ink density). Today I repeated the experiment on a P700 printer, which has ink density control. I printed three 255-step tables on clear film, using photo black ink and maximum Color Density (+50%). The first step table with no tint, the second with a red tint, and the third with a green tint. I applied the tints in the printer driver. For example, for the red tint...
I measured the step tables on the transparency using an XRite 361T transmission densitometer and plotted the densities against grayscale tone (K=0-100%). The results show that either of these tints completely removes the hockey stick.
The tint curves also have a steeper slope in the vicinity of K=0% (the other problem area). So I think I found a solution.
AdChoices