Skip to content

Argyll Chart Definition HOWTO

Jump to bottom
Kai-Uwe Behrmann edited this page Dec 22, 2017 · 1 revision

A brief HOWTO on the creation of Argyll chart definition files (.cht).

All descriptions here are based on the Wolf Faust's "C1" target. All descriptions are outlined for a Linux system, but should otherwise apply well for other systems as well after making some obvious changes.

The full description of the chart file definition can be found in the Argyll documentation. However, this entry is meant as a recipe to create a proper chart file from an arbitrary target. For a closer description of the fields in the BOXES section refer to the original Argyll documentation.

  • Obtain a good image of chart with scanner or camera (TIFF, no JPEG compression, no alpha channel)
  • Create a basic chart file sleleton using scanin -g 
 $ scanin -g myScan.tif myTarget.cht
  • When editing, keep a blank line after each section
The skeleton chart file we have obtained through this is in pixel units for all coordinate values given. Actually it can be done in any values. The scanin program goes and uses only the relative proportions of this file to apply them later to the acquired images of the test charts. All further coordinates mentioned therefore here are based on the generated skeleton in pixels.

Now, we need to edit/add the necessary sections for the chart file:

  • REF_ROTATION -- Keep this section tag with its value, as it is necessary to correct some of the values used in the definition for rotation angles
  • XLIST and YLIST -- Clean up all entries to the following:
    • All corners of diagram
    • Fiducial marks
    • All lines used in patch grid marks
      • When identifying which (of several) "ticks" to use, find one with a high value in the second column, and discard the ones with a low value
    • Adjust the integer after the XLIST and YLIST tags to reflect the number of entries following in that section
  • BOXES -- Add this section between the REF_ROTATION and the XLIST sections
    • Tweak the integer after the BOXES tag until it fits (for me number of patches + 1 worked, but others are seen in other chart files ...)
    • F -- fiducial marks
      (use top left, top right, and bottom right mark positions)
    • D -- diagnostic box
      (enclosing the whole "interesting" area of the chart, with measuring area and fiducial marks, but smaller than the box enclosing the total chart)
    • Y -- sequence boxes, array of patches
      (columns, then lines)
      • Coordinates in sequence boxes:
        • first the range of column labels
          (start to end, e. g. "01 22" for columns 1 through 22)
        • then the range of line labels
          (start to end, e. g. "A L" for lines "A" through "L")
        • width and height of the patch sizes of these sequence boxes
          (e. g. "36.8 36.8" for square patches with 36.8 pixel units in average size)
        • top left starting coordinates of the first patch in sequence
        • offset horizontally/vertically of next patches
          (if the patches are square with no gap next to each other this should be the same as the width/height pair for the patch size mentioned above)
    • X -- sequence boxes, array of patches (lines, then columns)
      • Definition the same way as the sequence box above. If the box contains a linear array (patches in one straight line only), the not needed dimension can be "dummied away" using the underscore ("_") as a place holder for the label range.
  • BOX_SHRINK -- "Shrinkage" of patch area (estimate something that's sensible in the coordinate units, e. g. 4.0 pixels)
    • Add this section right after the BOX_SHRINK section, still before the XLIST section
  • EXPECTED XYZ -- Add measurement values from vendor information (e. g. in XYZ colour space). In case of CIE LAB values, replace the "XYZ" by "LAB".
    • Add an integer declaring the number of defined measurement patches to follow in this section.
    • Add for each patch the index with the three colour coordinates. These can be obtained either through direct measurement with a spectrophotometer or from the manufacturer's target specs that are usually supplied along with the target. Pad the numbers (if necessary) to give "nice" labelling (e. g. A1 --> A01)
   A01         2.90    2.51    2.30

Test the created chart description using scanin with the used target image and description creating a diagnostic image with the boxed drawn in: 

 $ scanin -o -dionap myScan.tif myTarget.cht diag.tif

The image diag.tif can be viewed with any image viewer. It should present all properly aligned and identified patches of the target.

Clone this wiki locally