Tag Archives: color

ColorMunki – Review

See the earlier posting  ColorMunki – Features for a description of the ColorMunki. This section is on my experience with the device.

Ease of use

The ColorMunki itself and its software are generally easy to use. The device’s design are a bit of reminiscent of Apple products (the brochures states “Swiss Design” for whatever that’s worth) and there is even a white version called ColorMunki Design (as opposed to the black ColorMunki Photo). The software tries hard to hide the technicalities of color management. This is good for the many users interested in just getting the job over with – especially when you need to recalibrate a monitor every few weeks. It is a bit less suited for nerds or other control freaks who want to be certain that they are getting the best results possible (although v1.1 software does give a bit more control than the v1.0 software).

Nevertheless, some useability issues remain (although I downloaded the latest PC software, the hardware and firmware is a bit older, so there is a chance some hardware issues may have been fixed):

  • Sliding over glossy paper
    It is tricky to push or pull the device across glossy photo paper due to the extreme friction. A bit like trying to slide a rubber eraser over a glass table. It is best to push the device away, with your other hand (behind the device) holding the paper down.
  • Dial and software sometimes disagree
    The ColorMunki has a large dial to change between usage modes. The dial actually mechanically rotates the internal sensor to point down (profiling and spot measurements), up (ambient light measurement), 7:30 (device calibration), 10 o’clock (projector calibration). The software sees the dial position, but regularly looses track. This is easily worked around because the PC software what it thinks the position is. If it looses track, you move the back and forth a click.
  • User Interface quirks
    Similarly, the software guides you though a multi-step process: connect, calibrate, scan, etc. The software can get confused if you accidentally deviate from the standard path. A simple example: when installing the software, you can register (web site) and activate the product. The activation button is often disabled – e.g. because the device is not connected. You don’t get a message why the Activate button is disabled, and you can’t proceed until you figure it out.
  • Confirmation
    The designers focus on getting the user to do the steps they believe are needed for profiling. Fine – they know their color management. But no optional steps/checks are provided to demonstrate or prove that the profiling worked well: You don’t get to see the profiles as graphs/projections/3D shapes. There is no feature to easily see before/after prints (feed same sheet through printer twice, and show before/after as left/right). There is a before/after feature for the monitor, but the test image is small and limited compared to the similar feature on Spyder 3. You don’t quantitatively see how random colors are off from the previous- or ideal colors (the device can more or less objectively measure its own performance).
  • USB power consumption and hubs
    The device needs to be plugged into a USB port of the PC, or into a powered hub. Plugging it into an unpowered hub will result in an error message (unpowered USB 2 hubs support devices up to 100 mA). This is because the ColorMunki’s USB interface reports that it needs 500 mA (@ 5 Volt = 2.5 W). This number is presumably over-pessimistic for a device with an LED and presumably very little silicon. At 2.5 W, I expect to be able to operate a device like an iPhone with its most power-hungry features enabled. You won’t run into this problem if you plug the device directly into the PC, but I lost some time on this because a device that looked like a passive USB extender turned out to be a hub.

The above list is longish, and I think that te software still has rougher edges than a key competitor (Spyder 3). On the other hand, none of this is a big deal. The list is mainly intended to increase the chance that things are fixed.

Accuracy

In reviews, ColorMunki generally has a good reputation for accuracy. But one practical problem is that there is little you can do to check whether it is working well for your display/printer: you will generally get different colors (especially for the display) but this doesn’t tell you whether you made things better or worse.

Some attempts at checking the setup with my Samsung SyncMaster 206bw display and HP B9180 A3+ pigment printer:

  • I scanned a pad of shocking pink Post-Its and compared the color of the pad to the swatch color shown on the screen. This didn’t work well. There are, however, 2 serious technical flaws in this test:
    • Firstly the ambient light used to judge the Post-It will obviously affect its perceived color. Professional color specialists avoid this problem by using color booths: tabletop miniature rooms with calibrated lighting ( D50 or D65 color spectrum).
    • Secondly,  a Post-It color may be the worst possible thing to reproduce. It seems to have fluorescent properties: incoming light of one wavelength can result in light of a different wavelength. Fundamental color spaces do not account for this (exceptional) condition. Scanning of more standard printed material worked better.
  • An ICC profile viewer (free PerfX 3D Gamut Viewer) can be used to compare a new profile made using ColorMunki against a from the a monitor/printer/paper manufacturer or other source. The measured and downloaded profile pairs looked similar – but not identical – in shape for both my display and my printer. This confirms that the results are more or less OK. But it only (?) shows the extremes of the color gamut (surface of the L*a*b plot). And it doesn’t say which profile is better. Fancier commercial ICC profile viewing software is available from Imatest (Gamutvision) and CHROMiX (ColorThink).
  • An obvious other test is to do a before/after comparison for the screen. The screen had been previously calibrated using a borrowed Spyder3. The difference was visible (more or less a color temperature difference), but it is obviously tricky to say which is right. To complicate matters, you even can specify what color temperature you want the screen to be. For printer calibration, the printer was previously presumably calibrated in the factory and using a built-in self-calibration feature. The differences were clearly visible for parts of the color gamut (degree of yellowness of a light orange/brown). Again, the tricky part is which print was better.

ColorMunki – Features

I recently bought a (used) color calibration device: the ColorMunki Photo from X-Rite. Although it looks like an oversized tape measure, it is used to calibrate computer displays (either LCD or projector) and it can calibrate printer/ink/paper combinations. In either case, calibration is done by comparing the output colors against the intended color and by adjusting for the inevitable deviations within the computer. This kind of calibration is used heavily in the photography and graphics industries or by anyone else who needs accurate colors.

Display calibration

De ColorMunki calibrates the colors of an LCD monitor (or projector) by measuring the output colors while it controls the display using special software. It then generates unique correction data (an “ICC profile”) for compensating for display behavior within the graphics hardware in the computer. For an LCD monitor, the measurement device actually is hung facing and touching the LCD screen, while for an LCD projector the devices monitors the screen from a few meters away. In both cases, the device measures the colors of the light coming from the screen (this takes a few minutes and is largely automated).

Display calibration is generally done by serious photographers: if you don’t have a calibrated display, there is little point is worrying about exact colors of photos or prints.

Supplied ICC profiles for displays are seldom good enough because the display colors depend too much on brightness/contrast/color temperature settings. Furthermore displays (probably mainly the backlights) age, recalibration is needed every 1-4 weeks.

Printer calibration

Printer calibration is a bit more complex and done by fewer people. This is because a manufacturer will typically supply accurate ICC profiles for a printer for major (often its own) paper types. Because periodic recalibration is less essential, and because profiles for major printer/paper combinations can be found on the Internet, one can often do without. The ink also plays a major role in this, but people who worry about color will typically not want to complicate matters by using 3rd party inks.

Printer calibration again uses special software (supplied with the device) to print test charts with 5 strips of 10 color patches each. The ColorMunki measures by illuminating the patches using a built-in light source (presumably white LEDs) that it can in turn calibrate using a built-in reference target  or tile(which is not directly visible). You can a scan a strip of patches at a time by slowly pulling (works better than pushing) the sensor over each strip. The software will tell you whether or not the strip was scanned properly – if not you just repeat that strip.

As first original owner warned, this “pulling” over glossy paper does not go smoothly because the Teflon runners at the base of the device have a lot of friction when used with glossy paper. In fact, you can see some faint scratches on the photo after the scan. These do not impact the scan because the optical sensor is between the two runners. The sticky Teflon problem can be resolved by covering the Teflon with scotch tape or Post-It paper.

Compared to fancier printer calibration devices, this device uses fewer patches (a minimum of 2×50 – although you can add more if you want) and the stripwise scanning is probably unique: the alternative is to manually scan each patch or to use a small robot-like arm to position the sensor in two dimensions.

Projector calibration

The device can also calibrate a projector attached to a computer. This is similar to calibrating an LCD monitor except that the device needs to see the screen from a distance (equal to roughly the screen width). As the device will be nearer to the screen than the projector, you should avoid shadows or limit them to the edges of the screen. This is not a big problem because the “eye” of device can be adjusted to point upwards a bit.

Unless you project in te dark (like a movie theatre), the blackest black you can get with a projector is determined largely by the ambient lighting. I doubt whether you would need to recalibrate if the ambient lighting changes: the ColorMunki essentially just measures the obtainable lighting range (per color) and ensures that the intermediate values are sufficiently linear. Arguably the only impact of significant background lighting (providing it doesn’t change during calibration) is to increase measurement noise slightly.

Real world color extraction

The scanner can also scan the color of printwork not coming out of your own printer or even physical objects (paint, cloth, etc). This is probably more useful for designers who need to specify or match product colors, but may come in handy if you want to compare a photo to the object being photographed. If you like, you can consider the device a 1-pixel color calibrated camera. You can sample colors a few millimeters in diameter which the ColorMunki can be placed on (not a car across the street).

Software

As the software has recently (July 2009) been upgraded from version 1.0 to 1.1, I directly installed version 1.1. This provides a bit more technical control if you want that. But also gets rid of an annoying 3-computer activation limit. This 3-computer limit was mentioned as a drawback in many reviews, but was actually never enforced in any way.