I recently purchased an Xrite ColorMunki Photo, and have set it up to be used with ArgyllCMS and Osram Color Calculator. The TM-30 reports I’ve generated seem to be in range of what I could expect, but I wanted to ask how people using these tools ensure consistency/accuracy? I have my spectrometer placed on the floor beside my pc under my desk, and when I run a spotread the only light source on in the room is my monitor. Does that even matter considering ArgyllCMS requires a calibration before use? Any advice would be greatly appreciated!
Here are three reports of my Zebralight SC65C HI; levels L1, M1, H1.
Repeatability is easy to test, as for accuracy, can’t really know for sure without a fully known reference light. There is a partially known reference : an incandescent bulb (or any incandenscence, e.g. a candle), of known duv of 0 and CRI of 100. If the spectrometer is close to that it’s already a good indication of accuracy.
I thought the colormunki photo does calibration via an internal light source, which is why the sensor is rotated and unexposed to ambient light during calibration. I could be entirely wrong about the internal light source, but I think I am at least correct that the sensor is not exposed to ambient light during calibration.
Thank you for this, I’ve seen data of a candle using a spectrometer (one of your replies somewhere lol). I didn’t think to use that myself to check accuracy. I’ll test a candle tonight and take a look at the result. I’ll upload the report here for fun .
It would be interesting to compare tests from a nearly identical source, such as the same model of Maglite bulb powered at the same voltage, with the sensors placed the same distance from the source.
Yes. A steady fixed voltage is necessary for a reference test because the CCT will vary with the operating voltage of the bulb.
I’ve got benchtop power supplies, and various kinds of maglite bulbs. I could just test all of them at their rated voltage. Larger maglites have a spare bulb in the tailcap, so using the spare would avoid runtime hours having an effect on the results.