color calibration

 

It has been suggested that this article or section be merged into color management. (Discuss)

The aim of Color calibration is to adjust the colors of one output device to match that of another. The device that is to be calibrated is commonly known as calibration source; the device that serves as a comparison standard is commonly known as calibration target. Both target and source can be a Color space such as Adobe RGB or CMYK color space, a test print, color chart or material sample.

Information flow and output distortion

A computer program that sends a signal to the computer's graphic card in the form RGB (Red,Green,Blue) 255,0,0, signals only a device instruction, not a color itself. This instruction then causes the connected display to show Red to the maximum achievable brightness, while the Green and Blue components of the display remain dark. The resultant color being displayed, however, depends on two main factors:

• The phosphors or crystals actually producing a light that falls inside the red spectrum and
• the overall brightness of the color resulting in the desired color perception. (An extremely bright light source will always be seen as white, irrespective of spectral composition.)
  

Hence every output device will have its unique color signature, displaying a certain color according to manufacturing tolerances and material deterioration through use and age. If the output device is a printer, additional distorting factors are the qualities of a particular batch of paper and ink.

The conductive qualities and standards-compliance of connecting cables, circuitry and equipment can also alter the electrical signal at any stage in the signal flow. (A partially inserted VGA connector can result in a monochrome display, for example, as some pins are not connected.)

Color Perception

Color perception is subject to ambient light levels, and the ambient white point. (A red object looks black in blue light.) It is therefore not possible to achieve calibration that will be perceived evenly in different lighting conditions. The computer display and calibration target will have to be considered in controlled, predefined lighting conditions. Controlled lighting conditions such as D65 help to suppress the effect of metameric colors which would further complicate the issue.

Techniques and procedures

The most common form of calibration aims at adjusting monitors and printers for photographic reproduction. The aim is that a printed copy of a photograph appears identical in saturation and dynamic range to the source file on a computer display. This means that two independent calibrations need to be performed:

• The computer display needs to represent the colors of the image color space.
• The printer needs to match the computer display.

In the first stage, an external calibration device is attached flat to the display's surface, shielded from all ambient light. The calibration software sends a series of color signals to the display and compares the values that were actually sent against the readings from the calibration device. This establishes the current offsets in color display. Depending on the calibration software and type of monitor used, the software either creates a correction matrix (i.e. an ICC profile) for color values before being sent to the display, or gives instructions for altering the display's brightness/contrast and RGB values through the OSD. This tunes the display to reproduce parts of a desired color space. (It is technically and procedurally not possible to reproduce an exact color space on a computer display). The calibration target for this kind of calibration is that of print stock paper D65 at 120 cd/m².

In the second stage, the software sends a test print to the printer and compares the print result with the original file with the use of an external calibration device, similar to the display calibration. A calibration profile is necessary for each printer/paper combination.

Calibration Devices

Several colorimeter devices available for calibrating display monitors are listed here.

Shown prices (as for mid-2007) are for comparison only.

Device name	Manufacturer	Versions and features
Huey	GretagMacbeth	Standard ($90) — basic Pro ($130) — medium
Spyder	ColorVision	ColorPlus ($70–120) — basic
Spyder2	ColorVision	Express ($90) — basic Standard Suite ($180) — medium Plus ($250) — medium (just a software bonus) Pro Studio ($300) — advanced
Eye-One Display 2	GretagMacbeth	Standard ($260) — medium LT ($180) — basic
MonacoOptix XR (DTP94)	X-Rite	Standard ($300) — medium Pro ($450) — advanced

Note about manufacturers and vendors:

• GretagMacbeth was acquired by X-Rite in February 2006, but still sells products under own brand name.
• Original MonacoOPTIX was manufactured by Monaco Systems until 2003, when the company was acquired by X-Rite. The new product name is “MonacoOPTIX XR” (with “XR” meaning X-Rite).
• ColorVision is a subsidiary of Datacolor.
• Pantone was selling ColorVision products till 2006 (such as “Pantone ColorVision Spyder”), then switched to GretagMacbeth (“Pantone / gretagmacbeth Huey”).

Feature classifications:

• Basic — fixed gamma and color temperature (2.2/6500).
• Medium — selectable calibration tagets, even custom (Spyder2 Standard).
• Advanced — professional capabilities:
  • Characterization without calibration. You can calibrate the display to whatever target you want, even leave it as is.
  • Pure arbitrary gamma and color temperature (MonacoOPTIX XR Pro). You can define any tonal response curve, even that non-exponential of sRGB, by simply supplying a text file with numbers. You can specify target temperature not only by a correlated value (CCT), but directly by xy chromaticity coordinates.
  • Calibration verification and profile validation: reduced series of measurement to check whether profile still describes monitor characterisicts well enough; average ΔE difference is reported. While Spyder2 verifies against the same RGB-shades as used for calibration and profiling, MonacoOPTIX applies color-checking charts of different hues. MonacoOPTIX also tracks measured deviation and displays a trend diagram for estimating how often the display needs re-calibration and profiling.
  • Table-based profiles (MonacoOPTIX XR Pro) allow you to characterize very sophisticated devices. This is obligatory for printers and scanners, but for displays usually renders worse results than standard matrix-based profiles (could add 1 ΔE or more).
  • Measured tuning allows you to set up monitor brightness (white level) and bias (black level) to a desired numerical value, rather than with visual assessment. Actually, the software only informs you about current luminance level — it doesn't modify videocard's color look-up table (which is done for gamma correction). So if you monitor has no bias control, you can't raise black level without affecting white level (when controlling brightness or contrast).
  • Multi-monitor matching (MonacoOPTIX XR Pro). Despite of challenging name, this function just helps you to find “least common denominator” between several display profiles, so you can perform measured tuning of brightness range among all monitors being matched. No direct modification of chromaticity is done.
  • Projector calibration (Spyder2 Pro). Special holder is supplied to fix the sensor on a tripod.
  • Free measurement mode (Spyder2 Pro). You can measure what you want and see the color coordinates: XYZ and Yxy (cd/m²), L*u*v*, kelvins, etc. With standard versions of software you can't see these numbers, except for luminance and temperature in some cases.

Note that all those features (except tripod-holder, which is a piece of hardware) are implemented in software. So all versions of a given product share the same hardware — no reason for bying a pricy version if you don't need advanced features. But with Eye-One products you can't upgrade to a higher level by simply purchasing a license (i. e. software registration serial number), because software capabilities are unlocked by serial number of sensor itself.

Special features:

• Before and after pictures for comparing the result against uncalibrated state (Huey, Spyder2).
• Monitor selection in multi-monitor configurations is usually available only with Pro-versions.
• One-button calibration (Eye-One Display 2). Compatible monitors can be fully automatically calibrated through DDC/CI protocol. Eye-One software v3.5 supported only 8 monitor models (primarily high-class products), while in version 3.6 the compatibility list expanded to some 22 models. For a monitor to be supported a text file with capability description is required; eventhough not to hard to create manually, you need to know internal configuration names and ranges (looks like OSD menu specification).
• Continuous illumination adaptation (Huey). Works like internal illumination sensor in some monitors; adjusts luminance level only, not color temperature. This feature is controversial in context of constant color reproduction.

Measurement speed:

• Huey — fast. Only 50–60 seconds is required. 26 test patches are measured in 2–2.3 seconds each.
• Spyder ColorPlus — very slow.
• Spyder2 — slow: about 14 seconds with IR-buffle (for LCD and projectors) and about 6 seconds without (for CRT). With 67 patches to measure, total required time is 6.8 minutes for CRT and 15.5 minutes for LCD. Professional version allows skipping RGB-shades (3×16 patches) or gray-shades (16 patches), depending of technology of measured display, thus reducing total time to 2.2 minutes for CRT, 7 minutes for LCD and projector. Verification against 21 patches (16 gray + primaries) is done in 2.1 and 4.9 minutes respectively (2.2 minutes for DLP projectors).
• Eye-One Display 2 — moderate; 4.5 seconds per patch. Reading of 65 patches requires 4.9 minutes.
• MonacoOptix XR — fast; 3.4–3.7 seconds per patch. Matrix based profiles require 2 minutes with 35 patches (8×RGBW + 3 checks), while table-based measurement lasts up to 5–6 minutes (CRT/LCD). Verification with an assortment of special 24 patches takes only 1.3–1.5 minutes.

An important note about measurement precision. Until recent time, no official specification was published for any of the aforementioned product. For the moment (2007), it is not the case anymore, except for Huey (which incorporates the same technology as Eye-One Display 2) and Spyder ColorPlus (which is outdated). This specifications are not uniform in type of data being published, so can't be directly compared in a comprehensive manner. But they apparently show similar working ranges of luminance and similar error values. Head-to-head comparison of Spyder2Pro and OptixPro (the only ones that allow to see Yxy numbers) reveals that with luminance less than 1 cd/m² results can differ more than 1.8x between different sensor types, eventhough lower bound of working range is 0.02–0.05 cd/m² for both of them. So, whichever of them is more precise, a caution should be made if the results are used for other purposes rather than monitor calibration. All this products are entry-level colorimeters, not lab-grade measurement devices.

Reference: Entry-level colorimeters, an updated version of review for PC World Russia #4/2006 (in Russian)

External links

• Free online tools for calibrating monitor brightness, contrast, color depth and resolution.
• Online monitor calibration tool
• A discussion of some color calibration tools

Donate to Wikimedia