An other Open Source Colorimeter

Richard Hughes, the author of colord, developed in the recent months new hardware for measuring monitor colours. The ColorHug called device shall come at a relatively low price. It shall be useable for LCD/LED monitors providing input to calibration and profiling software. The most wide spread open source colour management system, which can create ICC profiles from colour measurements, is Argyll.

ColorHugThe author Richard Hughes states on his blog entry: “Existing hardware is proprietary and 100% closed, and my hardware has a GPL bootloader, GPL firmware image and GPL hardware schematics and PCBs”. The “100%” is a wrong marketing claim as Richard Hughes should know as Argyll user. However the new device fits nicely into a row with prior open source art in colorimeter hardware like the HCFR. The HCFR is supported in Argyll since some years now. To make the new ColorHug device functional, it would be great, if the hardware author could deliver a module instantly useable in Argyll.

What would now be interesting is to know, how the new device will compare with pre existing ones, being them proprietary or open source licensed hardware. The author gave a hint about speed. But speed is only one property useable to reduce noise in dark readings. Much more interesting is colour accuracy.

What is colour accuracy and why is it so important for a colorimeter like the HCFR or the new ColorHug? Colorimeter devices suffer almost all from a difference to the ideal colour reception of human eye, especially the cheaper ones. Only spectrometers can compensate better for that effect of non perfect filters in front of the actual light sensors, but expose other disadvantages. Colorimeter devices, which perform close to human sensibility, are usual expensive. Some are even more expensive than colour spectrometers. Colorimeter manufacturers use a common trick and put a correction matrix inside the device, which shall compensate for the difference between the sensitivity of human eyes and the colorimeter. But many users complained not to be able to get good results despite. This is easily understandable, as monitors emit light with very different spectral characteristics, which do not match the used filter in the colorimeter and its matrix. One approach to get better results is to use a per monitor model compensation matrix. Fortunately Argyll has implemented compensation matrices in one of its recent releases. The requirement for this approach to work is, that the data base needs input data from users.

18 thoughts on “An other Open Source Colorimeter

  1. “An other Open Source Colorimeter”? There is no other open source colorimeter. Therefore ColorHug cannot be either “an other”, or “another”.

  2. The HCFR page that is linked to provides zero info on source code of the firmware, the schematics page is in French and as far as I can tell only provides a PDF. Neither there is a link to source code of the software. Prior open source art? Really? Well, if you wish to think so.

    • I prefere the english PCB PDF from the HCFR site, as it is instantly readable.
      Their site states:
      “This software is distributed under FSF. GNU GPL license.
      Sources code will be available on demand, ask authors.”
      There is a serious research missed on your and Richard Hughes side. The bold statement holds not true.

      • Dude, I never said there has never been any open source color hardware available before. I never stated that the ColorHug was in any way the the first open source attempt at this kind of thing.

        What I do know is you can’t actually buy the HCFR. You can’t get any kind of support, and a lot of the HCFR hardware info just isn’t available or accurate.

        • Richard Hughes: “Existing hardware is proprietary and 100% closed, and my hardware has a GPL bootloader, GPL firmware image and GPL hardware schematics and PCBs”

          Richard Hughes: I never said there has never been any open source color hardware available before.

          How can “proprietary and 100% closed” hardware be open source?

  3. A simple question: do you understand the difference between software and firmware?

    • firmware is not software? Can you then pls fix :P

      In any case – OK, so Richard wasn’t technically entirely correct when he stated the colorhug thingy is the first FOSS color meter but at least it might be easier available than the other one. Why bother fighting about this? Just make sure the sodding thing works properly with the color management stuff under Linux…

  4. Pingback: Links 20/11/2011: GNU/Linux in Tamil Nadu, Flex Donated to Apache | Techrights

  5. Getting away from which was first or more open or whatever … how well does the ColorHug work? What sort of black level can it measure accurately, what is the inter-measurement and inter-instrument repeatability, that sort of thing.

    Is the spectral response curve data for the three filters available, and what is the response curve of the sensor behind it?

    (In the interests of avoiding fights I refrained from asking why a UK private limited company is selling a ColorHug and not a ColourHug).

    • The ColorHug works about as well as a Huey, from my testing. It’s obviously not in the same league as a ColorMunki, as it’s 1/20th the price. Each ColorHug device is calibrated individually from a 836 measurement set, although so far the sensor response seem to be pretty constant between devices. I’m waiting for the large sensor delivery today before I can measure the differences for a larger number of sensors.

      I’ve uploaded the frequency response here: — which is a combination of the sensor sensitivity data multiplied by the IR filter data and smoothed a bit for simplicity. The factory calibration matrix converts a single set of primaries to XYZ color, and then the device has 63 slots to fill with custom ccmx correction matrices. This gives the device an acceptable performance when the primaries don’t match my idealized sRGB calibration colors.

      It’s also called the ColorHug, as after developing color software for many years, it’s actually harder for me to type colour than color now. :)

  6. Richard, thanks for the rapid response.

    I saw that graph, I was looking though for the actual data not just a raster image of a graph. Perhaps it could be added to the git repo?

    With wide-gamut ccfl, ‘white’ (blue with orange filter) led, rgb led and normal ccfl backlights all in comon use, support for multiple ccmx matrices is a very welcome feature, great to see that.

    There seems to be more than one device called colormunki: the ColorMunki Display which Amazon US sells for USD 139 and the ColorMunki Photo which Amazon US sells for USD 449. The former seems to be a colorimetyer, similar to the i1Display which I have, and the latter seems to be a spectrophotometer (sice it can do printer profiling).

    Anyway I have preordered a ColorHug, and further discussion is probably better on the dedicated list for it which have joined.

Comments are closed.