Introduction to Multicharts:  Measure color accuracy, tonal response, and noise from a large variety of test charts

Imatest™ Multicharts analyzes images of a great many standard and custom color test charts for color accuracy, tonal response, and noise using a highly interactive user interface. It measures white balance and color response for a wide range of lighting conditions and scenes. It can also display the tonal response of monochrome charts and calculate a color correction matrix (Imatest Master-only). Multicharts currently supports:

Q13 grayscale step chart

Sources of charts are listed on Using Multicharts.

You can select either standard chart reference values or you can read in values from data files (Imatest Master only). Values for the two ColorCheckers have been supplied courtesy of X-Rite. You can select between six image file color spaces: sRGB, Adobe RGB (1998), Wide Gamut RGB, ProPhoto RGB, Apple RGB, or ColorMatch. Danny Pascale/Babelcolor’s page on the ColorChecker contains nearly everything you want to know about the chart. Ian Lyons has a nice description of IT8.7 charts.

Before running Multicharts, photograph the chart, taking care to illuminate the chart evenly (±5%) and avoid glare. For testing white balance, you can photograph the chart in a scene (the mini ColorChecker is especially suitable). Then click on Multicharts in the main Imatest window. To load the image file, click the appropriate chart type in the box below Read image file. Chart type: Crop it, then enter any needed data in the input dialog box.

Differences between Colorcheck and Multicharts

  • Multicharts can analyze a great many standard and custom color and grayscale charts in addition to the 24-patch Colorchecker.
  • Multicharts is highly interactive. You can choose among a large number of displays and options after the the chart image has been entered.

Example: the 24-patch ColorChecker, shown with the 2D a*b* plot

The left image shows the 2D a*b* display. Ideal (square) and camera (circle) {a*, b*} patch values are shown. The Chart view box allows you to select among several displays..
The upper right image shows the color chart. One of several displays can be selected with the buttons just below the image: Ideal the ideal image (as is; unmapped), Id>M the ideal image mapped to sRGB color space (which is close to most monitor displays), Input the input (camera) image unmapped, and In>M the input (camera) image mapped to sRGB.

Color Charts module, 24-patch ColorChecker
2D a*b * display for ColorChecker

The lower left region shows color differences, using several different metrics.
The lower right region is the display control and option area.

The available displays, selected by the Display box, include

  • Pseudocolor color difference. Shows the difference between the ideal and input color patches using any of several metrics. Probe is available for this display.
  • 3D color difference.
  • Split colors: ideal/input. Shows the patches split so that the ideal and input values are displayed in the upper left and the lower right or each patch region, respectively. Probe is available for this display.
  • xy chromaticity.
  • u’v’ chromaticity.
  • Black & White density.
  • 3D L*a*b*. Shown on the right. L*a*b* coordinates can be difficult to visualize when viewed from a single angle. It can be rotated (automatically by pressing the A-rotate button) for improved visualization.
  • EXIF data and Color matrix.  Shows EXIF data if available and the color correction matrix if it has been calculated.
  • Noise analysis for charts with large enough patches, including chroma and RAW sensor noise.
3D L*a*b* display
3D L*a*b* display for ColorChecker

Examples of different charts and displays

IT8.7 chart, split view
Split display for IT8.7

This Multicharts split display is shown above for a reflective IT8.7 chart, with the Probe turned on. In each patch, the upper left portion is the ideal (reference) value, converted to sRGB for screen display, while the lower right is the input value, also converted to sRGB for screen display. The image is synthesized.

The Probe is turned on by checking the Probe box on the right. When the probe is on, you can probe any patch in either image. The probe data (for the individual patch), shown below the split display, includes an image of the ideal and input patch colors, ideal and input RGB and L*a*b* values, and several ΔE* and ΔC* color difference metrics.  ΔE*ab, the geometric distance in L*a*b* space, is the most familiar, but ΔE*94, which is lower for saturation differences in saturated colors, may be a better measure of visual color difference. It’s easy to correlate ΔE* metrics with visual color differences using this display. None of the metrics are perfect! The probe is turned off by clicking outside the images.

The pseudocolor display is shown on the right. Differences can be displayed for several color metrics: ΔE*ab, ΔC*ab, ΔE*94, ΔC*94, ΔE*CMC, ΔC*CMC, ΔL*, ΔChroma, Δ|Hue distance| , Δ(Hue angle) , Chroma (input) and Chroma (output). Several color maps are available.

Pseudocolor display
Pseudocolor display for IT8.7

IT8.7 charts must be used with reference files, which are available from the manufacturer’s websites.

IT8 xy Chromaticity display
xy Chromaticity display for IT8.7

The xy Chromaticity diagram for the IT8.7 chart is of special interest. To avoid clutter it doesn’t display all the patches (this is also the case for the ColorChecker SG and the CMP DC003). Instead it displays values for one of five zones in the chart: columns 13-19 (primaries), 1-4 (dark), 5-8 (mid), 9-12 (dark), and 20-22 (miscellaneous— different for different manufacturer’s charts). The primary region (columns 13-19, shown above) is the most interesting since it shows the response to the subtractive and additive primaries, and gives some indication (though far from perfect) of the color gamut of the device + color space under measurement.

B&W display for Q-13
B&W plot: Log(Pixels) vs. Input density for Q-13

The final example of the tour shows the grayscale response curve, in this case Log10(Pixel level) vs. Input density. Gamma is the average slope of this curve for the lighter regions. This plot is also available for all of the color charts because all of them contain grayscale regions. Several additional B&W curves are available.

Log(pixels vs. input density) The slope in the region where input density is between about 0.1 and 1.0 is gamma. Curves for the R, G, and B channels are shown in addition to Luminance.
Output vs. input density The slope is ideally 1. Density is proportional to -Log10(Luminance).
Output vs. input luminance Shown on a linear scale.
Output vs. input L* a* b* c* Uses CIELAB values instead of density or (linear) luminance. Includes color values: 10x a* (green-red), b* (blue -yellow), and c* (chrominance: (a*2 + b*2 )1/2


You can learn more about Multicharts from Using Multicharts.