Case Study: Characterizing color negative film that does not match one of the built in types

Our user Daniel von Mitschke recently posted a question on Flickr's ColorNeg / ColorPerfect user group which basically was a follow up question to our earlier general discussion of dealing with cyan color casts in the highlights of negative conversions. Mr. von Mitschke provided some example files of his linear 6x7 negative scans and kindly permitted us to use his photos in the following step by step description of the image editing process. For that we'd like to thank him cordially. This is the first of two pages detailing what I did with his images. It is intended to explain ColorPerfect's FilmType / SubType / FilmGamma characterization system a bit more detailed than before and by using another example image. Please note that the results presented may well be improvable by finding an even better characterization. Also note that all choices made after characterizing the negative are in the realm of artistic preference. Different choices would have led to different images.

The photos have been shot on Kodak Portra 160 NC film and were digitized using Silverfast on an Epson V750 scanner. Getting linear scans with Silverfast is rather straight forward and has been excluded as the source of the color cast issue with reasonable certainty.

About the negative film characterizations included in ColorPerfect

While ColorPerfect does include characterizations for a vast number of color negative films from the last decades it is important to understand that all of these have been derived from data published by the films' manufacturers. For some film types several data sheets exist for different generations of the film. We have generally included all data we could get hold of. In case of multiple entries for the same type of film the characterizations' names either contain elements in square brackets - being specific codes that are expected to be printed on the negatives' edges - or elements in curly braces - being properties that can't be determined from the negative itself but need to be known, like specific development conditions. Often one of the included characterizations is a reasonably good match for actual rolls of film of the respective film type but sometimes none is. For one mismatches occur because the film's processing has an influence on the actual negative gammas - with the variables being development time, temperature and agitation - and for another there may well be fluctuations between different batches of the same type of film.

Trying to process a sample negative using the built in characterization for Portra 160 NC

Let's start by opening the linear scan in Photoshop. You can download a downsized version of the original to follow along if you like. Normally we'd first assign our target color space to the scan but let's use sRGB for our example which has already been assigned to the file provided.

 

After loading the image we're ready to launch ColorPerfect. On the plug-in's start panel choose ColorNeg mode, make sure that the G/L button indicates "L" for linear input and set up Gamma C to match the chosen color space. For sRGB we have to select the according entry from the Gamma C list in order to utilize the correct tone reproduction curve. It is not truly a Gamma in this case but that technical detail makes no difference as long as you choose the correct entry.

Next let's select ColorPerfect's built in characterization of the film. To do so we choose Kodak from the maker list and Portra 160NC {2008} from the film list. An automatic initial color balancing operation will be performed upon our selection. That automatism uses physical relations in the brightest tones of the positive image and assumes that these tones relate to the light source of the scene. In some cases like a close up photo of a colorful flower this would fail but for the sky lit scene at hand it should lead to close to reasonable colors to begin with - which it obviously doesn't do here. The built in characterization does not match the scan at hand. That could be the case because of a number of things as explained above. For example the film might have been processed differently than at Kodak's reference conditions, it might be from another batch with different properties etc.

Using the FilmType / SubType / FilmGamma system

Be that as it may we'll have to deal with it and one way of trying to do that is attempting to find a suitable characterization by visual cues using ColorPerfect's FilmType / SubType / FilmGamma system. For that to succeed it is important to choose a good spot to white balance the image by gray clicking prior to starting the characterization process. The object in the scene chosen for this should be of a middle gray and should be illuminated by the scene's main lighting. It should not be in some very shaded region or in one with illumination that differs from the rest of the scene. The right side of the steel base of the gas station's price pole in the example image seems to be a reasonable choice. The same probably applies to several other surfaces in the scene. We right-click on the general region we prefer and then switch to 100% magnification. There we click several points in the image that we deem to be of a neutral hue until we have reached what we believe to be a suitable initial color balance. This really is a bit difficult to assess because without a proper characterization color integrity does not yet exist and the other colors of the scene won't follow the gray click to confirm our choice. We will maintain color casts in the highlights and depths which only a correct characterization of the film can remedy. Once we have settled on a color balance we want to try going on with we can optionally set saturation to an unnaturally high level like 200%. This will sometimes help us notice even slightly off colors while using the FilmType / SubType system. You can try the following with and without doing that.

The built in characterization we selected before has already been converted into the three parts of that system: a FilmType setting, a SubType setting and a FilmGamma setting. The latter becomes available instead of the artistic Gamma setting when SubType is active. This will be our starting point. It differs depending on which built in or custom characterization has been selected before getting started with the characterization process. Our starting point in terms of FilmType / SubType is [030.949] because of our selection of Kodak Portra 160NC {2008}.

To start adjusting the values we click into the FilmType / SubType input field. On the PC we can then use the scroll bar to slowly scroll over the entire range possible to see if we can find a setting where there will be less obvious color cast problems. On the Mac we have not yet managed to get the scrollbar to give a live preview within a plug-in for which we apologize. There you'll have to repeatedly click above or below the slider to cover the range or drag it short distances at a time until we do. Doing that in this example seems to lead us nowhere though. The entire range produces characterizations that in combination with our extreme saturation level are far off from what would seem natural color. In scrolling the entire range with some experience one could already notice two complimentary color casts that are adjacent on the slider's range. Around FilmType 330 we have a magenta cast in the sky and around FilmType 150 we have a green cast. Somewhere in between one would expect a turning point where the sky is neither green nor magenta.

With our starting point however it's hard to determine this spot. Situations like this can occur if the SubType setting derived from the starting point is very high or very low. In our example the SubType setting is too high for us to reasonably work with. So let's click on the FilmType button to switch over to SubType scrolling and cut the setting for SubType roughly in half. Let's set it to 450 and switch back to FilmType mode. In scrolling the entire range of the slider again we notice that the settings of FilmType 150 and Film Type 330 I pointed out before have become less extreme. Now we can select what seems to be the turning point from one complimentary cast to the other. A setting of around 255 seemed to me to produce the best result. In all of the other settings the color casts remained more obvious. This does not mean that this is the best possible FilmType but we choose it as it stands out from the rest. Staying at 255 or whatever other setting you like better we switch to SubType scrolling. Scrolling SubType up from the value we defined before takes us to one kind of extreme setting and scrolling it down takes us to the other extreme. Again we aim to find the point between the two where the scene looks most natural to us. I chose SubType 333.

Usually this entails switching back and forth a few times and trying to enhance the settings. I omitted that process in this description and used the settings I came up with in such a process in the write up above. The fine tuning generally is somewhat subjective and also depends on the quality of the initial color balancing. Any characterization produced this way will ideally have to be checked with more than one image from the same film and will probably have to be fine tuned a bit more - or even considerably - in doing that. We're at FilmType / SubType [255.333] now and it's time to look at the third component of the system: FilmGamma. It is probably the most difficult to set precisely without a known grayscale. I set it to 1.1 here and nudged FilmType a bit more in the process. FilmType is now 260.

At this point we have successfully established what looks to be one plausible characterization of our image source and can reset saturation to 100% in case we had boosted it before. Enter a name into the film list, click the Add/Chg button to add it to the film list for keeping. Note that we must then save the list in order to keep this state of it for future use. We can then always reload these values to use them or to adjust them further and then to save them under the same or a different name.

Artistic adjustments

We can now get on to the artistic adjustments. The following is very much more a matter of preference than the above. I first removed a considerable bit of black to brighten the image until the highlight clipping readout showed a value of around 30% for the entire image. The numeric value is of no consequence. I just specify those for you to be able to follow what I did. This negative shot on an overcast day recorded more information than can be printed in a positive image without special means. Why that is so is discussed on our page on the handling of highlights in negative conversions.

I then removed some white until the shadow clipping readout showed a value of around 7%. This is purely about visual choices. Don't let fear of clipping fool you. You are editing the image - not the histogram. I set shadow compression to 4 afterwards to preserve texture in the darkest tones. Such a high setting is not always preferable. It really depends on the image and I advise to look at the darkest regions at 100% magnification to see which setting is best. Simply right click on the point you want to zoom to and set the zoom level. While zoomed in you can always right click any point in the preview image and let that point become the center of the preview. By clicking repeatedly you can easily pan the image.

Let's now switch to the zone panel to apply a zone adjustment suitable to keep more texture in the sky than we would with highlight compression alone. I fixed zone 80 by activating it without changing the input zone and then let zone 110 become zone 100. Now we're ready to set up highlight compression. I chose 0.5 stops and a compression range of 240. To add a bit of contrast to the image I let zone 13 become zone 10. The image was not as saturated as I would have liked it so I set saturation to 115% and then exited ColorPerfect by pressing O.k.

In editing images like this one that do have a lot of detail in very bright regions one often tends to let the image stay too dark in fear of sacrificing too much detail in those brightest regions. Another easy way to work around that with more freedom than using highlight compression and zones like we did is making a quick second conversion directly after the first and blending the two in Photoshop. For that we restart ColorPerfect on the same linear scan or probably a second layer where we copied it to before and check the settings on the start panel. If things are at default settings they should still be correct. Then we'd press the "initial => previous" button to restore all of our previous editing choices. Now we can create a second version within seconds and can then combine the two versions in Photoshop by using the eraser, layer opacity, the fade command, the history brush etc. Using some of those tools I came up with this version of the image. Working with two more photos from the same film showed that the characterization we settled on above does not work as well for the other photos. A characterization I found that seems to work better with all of them is FilmType / SubType [280.207] with FilmGamma 1.25. Here is the resulting image with similar settings to the above. Again trying to improve this a bit further by combining two conversions etc. I made this final version. Comparing it to the first positive image we saw using the built in characterization shows that we have come quite a long way. It usually is easy enough to find a setting that looks better than what you have if things don't work out to begin with but finding ideal settings this way can require great care. Once the characterization has been established we can produce various versions of an image according to our preferences. E.g. I could easily have made the final image look like this using a simple but more elaborate adjustment to the underlying grayscale of the image.