Scanning Negatives using Vuescan, part 2: color profiling with IT8
 

0. Introduction - The Problem

The methods introduced in part 1 have one weakness: they assume that adjustments to white point, black point and gamma are sufficient to obtain realistic colors. Unfortunately this is not always the case. Quite often the color cast is different for shadows and for highlights.

We could obviously use more than a single gray point or otherwise elaborate on the methods. However, the easier way is to use an IT8 target with an appropriate software.

Part 2 of this tutorial shall show you how to:
- set up Vuescan so that the scans are appropriate for further treatment
- create profiles with the lcms profiler
- use your profiles

This tutorial has its limitations:
- it is not a manual for Vuescan or the lcms profiler
- it does not explain color management in general
- it is not suitable for beginners
 

1. Prerequisites

On top of the requirements mentioned in part 1, I assume:
- that you have understood part 1
- that you have an IT8 target (see http://targets.coloraid.de (I do not know of any other vendor). You might be able to get the same results with a normal, reflective target, but I did not try that.
- that you have downloaded and installed the free little color management (lcms) profiler from http://www.littlecms.com/profilers.htm
 

2. Why Do Profiling?

Color casts are often different in light and dark parts of the same image. Scanning a gray wedge might give you an image that is too red in the dark parts and too
green in the light parts. To correct this, you would need to apply a s-shaped curve to the red channel.

Correcting errors like this visually can be very frustrating. Any improvement in one place usually comes with a degradation in another.

Using an IT8 target has a big advantage: The colors on the target were measured with high accuracy and you get a file with the results along with the target. Even if the gray on the target is not 100% neutral, the file will tell exactly what color it is.

The lcms profiler is the perfect analyzer tool. It outputs ICC profiles. For those of you using non-ICC-aware picture editing software, Marty offers two little helper programs which can apply ICC profiles to TIFF and JPEG graphics.
 

2.1 Why use lcms (and why not Vuescan)?

Vuescan Pro comes with ICC profiling capability. Unfortunately, the resulting profile only contains a so-called conversion matrix.

Using the conversion matrix, Vuescan can correct the gamma for each of the color channels. The correction is close to what you could do by adjusting RGB brightness. It can not achieve s-shaped correction curves.

Ed Hamrick points out that Vuescan profiles can be used for different CCD exposures. Since color negative material can always be scanned at a single CCD exposure because of its low density, we can ignore this.

Vuescan calibrates using raw scanner data. That way, it always calibrates using the complete contrast range. However, an IT8 target will never need the full contrast range of any film, and calibrating that way will lead to errors in light or dark parts of the image.

lcms can generate a complex profile, consisting of a matrix or "tone reproduction curve" (TRC) and look-up tables. These profiles can be used to correct fairly complex color casts.

lcms needs an input file with a contrast range not exceeding that of the used IT8 target.
 

3. Taking pictures of the target and scanning them

The lcms profiler works better if the picture is as close to the reference as possible. It will even refuse pictures that are too dark or otherwise different. This, according to Wolf Faust, is a general problem of the ICC standard.

If you follow the steps from part 1, section 3.2, you will have a picture that is perfectly suited for the lcms profiler.

The picture has to be saved without own color space or profile. Make sure you select 'Device RGB' as the 'Output color space' on the Color tab and deselect 'TIFF profile' on the Output tab.
 

4. Profiling

The lcms profiler comes as a collection of programs, each doing a specific step of the process.
 

4.1 Measuring

We are going to use qtMeasurementTool.exe. The program can not read TIFF images, so we have to convert them. I prefer tiff2png (on
<http://www.libpng.org/pub/png/apps/tiff2png.html>) which can be used as the 'external viewer' in Vuescan.

Once you have started qtMeasurementTool.exe, go to the Options tab. Choose '22 column picker' under 'Pick template' and leave the rest of the settings at their default. (You can reset all of the settings by deleting the file lcmsmt.cfg before you launch the tool.)

Go to the Image tab and load the image using 'Load image'. Use the left mouse button to drag the grid to the right position over the image. Use the right mouse button to fix the lower right corner. The green fields should be completely inside the patches (the different regions of the target).

Click 'Pick' to make the tool read the colors values and display them on the Raw edit tab. Select 'Save IT8 Sheet' to save them into a file. Make sure you pick a meaningful name, and make the suffix '.it8'.
 

4.2. Creating the profile

We use qtScannerProfiler.exe to create profiles. Marti offers a different version on his web site (qtScannerProfiler1.exe), which is supposed to be specially suited for profiling negative film, but I wasn't quiet able to see the difference between the two so far.

It is a good idea to organize the data that comes with the target in the same manner as the profiler does, because it shows you only the description and date of the file, not the number of the target.

On the 'Scanner and Target' tab use 'Select target vendor & type' to choose your directory. Use 'Measurement' to select the .it8 file you just created, then choose a meaningful name for the profile under 'Output profile file'.

Do not forget to go to the 'Profile Identification' tab and type a short description! Your image editing software may show you this line later (Photoshop does).

Hit 'GO!' to start the actual process. The profiler will show you some status reports on the 'Progress' tab. Finally it will switch to the 'Results report' tab and show you the results per color patch.
 

4.3 Using the profile

You can visualize the profile using qtProfileChecker.exe. However, it makes more sense to apply it to your scan of the target!

You have to install the profile first. On recent versions of Windows, right click the profile, then select "Install" from the pop-up. You might have to restart your editing software for the change to take effect.

You can now load the target scan into Photoshop and apply the new profile. The new profile will show up in the list with the description you gave it.

If your image editing software does not know about ICC, you can apply the profile using tifficc. tifficc is a command line tool, so it can be used in batch files or added to the context menu. The syntax is tifficc -t0 -n -i<your profile> <input image> <output image>

tifficc uses sRGB per default. You can specify any other color space with the -o<you color space> option.

Example:

tifficc -t0 -n -iFilm01.icc -oAdobeRGB1998.icc img01.tif img01_c.tif

This command applies the Film01.icc profile to img01.tif, converts the result to the Adobe RGB color space and saves the result as img01_c.tif. All of the files must unfortunately be in the same directory, and <input image> and <output image> must not be identical.
 

4.4 Workflow

Having done all this, you should end up with a table containing one entry for each shot of the exposure bracketing done in part 1 paragraph 3.2.1. The entry contains the values from the Color tab and the name of the corresponding profile.

You should also note down the values under RGB exposure and Infrared exposure as well as the Film base color.

The actual workflow goes like this:

- Scan film with given exposure time (to raw file, if you like)
- On the Input tab, check 'Lock film base color' and 'Lock image color'
- On the Color tab, set 'Film base color', 'White point', 'Black point', 'RGB brightness', and 'Brightness' and set 'Output color space' to 'device RGB'.
- Scan (from raw file if you have produced one in step 1) and save (preferably as Tiff).
- Apply the corresponding profile. You have the choice: use TiffICC or load the scan into your image processing software and apply the profile there.

If you have done exposure bracketing and generated sets of parameters and a profile for each of them, try the corresponding profile first. If you don't like the shadows, try using the parameters and profile of the underexposed target. If the lights are not ok, try parameters and profile of the overexposed target.

CN-film handles overexposure better then underexposure, which means that you should try to get the shadows right when taking the photo.

5. Possible Problems

- As mentioned in part 1, colors vary from lab to lab. Film base color is often the only means of judging how much the colors are off. You can try to leave all other parameters as they are and to only recalculate film base color. If this does not help, you have to repeat the whole procedure.

- The color of the light on the target is the most critical issue. It is unfortunately very difficult to judge the color temperature of light. If you do not have the equipment to measure light temperature, you have to rely on your judgment alone. Early afternoon sun light seems to be most suitable. Standard 'daylight' lamps may not fulfill
the requirements. They spend a daylight comparable spectrum but still contain bands which may react different on chemical film as on human eyes (where they are optimized for).

- Each profile works with a given set of parameters and only with this set. You have to check 'Lock image color' to switch of all of Vuescans automatic features. Those could change white and/or black point leading to unforeseeable results.

- The target areas must fit within the boundaries of the areas of the measurementTool. If your image of the target is distorted, you should try to correct it using your image processing software. Be careful not to change its color space!