How can I create a camera color profile from an IT8 target on Linux for use with Darktable?

Asked 3/2/2015

3 views

2 answers

0

I use Darktable to convert CR2 files to TIFF/JPEG, and I’d like more accurate color than white balance alone gives me. I understand the basic idea as: buy a physical IT8 target with its reference patch values, photograph it, and use that shot to build a camera profile.

A few things I’m unsure about:

  • Do I need to photograph the IT8 target every time the lighting changes, like I would use a grey card for white balance?
  • On Linux, what software should I use to turn a photographed IT8 target into a camera profile?
  • How would that profile fit into a Darktable workflow?

If there’s a standard Linux workflow or a good tutorial approach for this, I’d like to understand the steps.

Originally by Photography Stack Exchange contributor. Source · Licensed CC BY-SA 4.0

Photography Stack Exchange contributor

11y ago

2 Answers

2

First, as to whether you'll want to shoot an IT8 target as often as you would a white balance target, the answer is "probably not". Getting a good IT8 shot involves significantly more consideration than shooting a gray card, and once you have a good profile made with a good full-spectrum light source, it will be perfectly usable for photos taken under most lighting conditions, adjusting only for white balance. If you have, say, an LED or fluorescent light source that you use often for important pictures, you might want to create a specific profile for that case.

As to what software, you'll probably be using the ArgyllCMS software tools to process an image file carefully prepared in your choice of RAW developers. Specifically, the scanin tool will attempt to recognize the layout of color swatches in the target image and read their color values into a data (*.ti3) file, and the colorprof tool will then attempt to generate a profile that maps those values to the nominal chart values with a minimal degree of error.

As to the actual process... it's complicated. The procedure can be as simple as in the link provided by @chili555 above, but somewhat better results can be acheived with some extra effort during the capturing as well as during the processing. The short version is that you'll produce a minimally processed TIFF from a carefully exposed RAW shot, in linear color space, which is white balanced for the light source and exposure compensated to match the documented tonal range of the IT8 card. The linked articles are a good starting point for diving in as deep as you're comfortable with.

As to personal experience, I found that the first profile I generated using the simpler approach was a substantial improvement over the stock profile provided in darktable (which, from what I understand, is drawn from the same source used by most other software, i.e. Adobe). The results of doing the extra work for a "well behaved" profile were subtler, but corrected some troubling issues related to color neutral values going into the profile not coming out neutral (using the spot white balance on a gray card, for instance, would not produce a gray output that was RGB neutral when sampled). My conclusion is that profiling is worthwhile if you're not happy with the results you're getting from software-provided profiles, but the results can vary depending on the effort invested.

Originally by user37649. Source · Licensed CC BY-SA 4.0

user37649

10y ago

0

AI Answer

Generated from our catalog & community — verify before relying on it.

You usually do not need to shoot an IT8 target every time lighting changes. Unlike a gray card for white balance, a camera profile is typically made carefully once under a good full-spectrum light source, then reused for most work with white balance adjusted separately. If you often shoot under a specific problematic light source, such as some LED or fluorescent lighting, making an additional profile for that setup can help.

On Linux, the common toolchain is ArgyllCMS. The usual approach is:

  1. Photograph the IT8 target carefully and evenly lit.
  2. Prepare the image in your RAW developer without unwanted color-altering adjustments.
  3. Use ArgyllCMS tools—especially scanin—to recognize the chart patches and compare them with the target’s supplied reference data.
  4. Generate a camera ICC profile from that measurement data.
  5. Use that profile in your color-managed workflow, including Darktable where applicable.

The key point is that profiling is more demanding than white balance: chart shot quality matters a lot. Make one good profile under a reliable light source, then use white balance for normal scene-to-scene variation.

UniqueBot

AI

11y ago

Your Answer