Firstly, thank you for the replies - many questions i had (and then subsequently got since writing this post) have been answered.
Hi Damien,
Do you shoot jpg instead of raw? Only then the preset wb of the camera will put a weight on the balance of the Planckian locus. If one shoots raw, then this can be adjusted using the "tint" (green <-> magenta) slider in the raw converter. But I guess I am stating the obvious to you, lol.
Cheers,
I am shooting jpg - this information is good as it confirms something I suspected from my data.
Hi Damien,
Which RGB values do you use, at what stage of Raw conversion? The output of the Raw conversion process is subject to a lot of data 'massaging' to produce a pictorially pleasing compromise. One of the important ones is White Balancing which can produce results on either side of the Planckian locus in a chromaticity diagram, and on many positions along it.
For a more reliable measurement one should use a better instrument, a (calibrated) spectrophotometer, which takes a direct reading of the spectral composition (which will also take atmospheric effects, e.g. pollution, into consideration).
Cheers,
Bart
For my Masters, we are deliberately looking at a DSC as the tool of choice. The camera is stated as being sRGB - thankfulyl, I have found much literature to convert from device dependent sRGB to CIE XYZ (and into the many color spaces and metrics)
Hi, Damien,
You speak of a high degree of correlation of the data, but you don't say with what. But I take it from your follow-on comments that you must mean a general correlation with the Planckian locus.
But (as I'll discuss shortly) I don't find it at all curious that that correlation deteriorates (in the purple direction) toward the "higher temperature" end of the locus. What I find curious is that the correlation appears at all!
The obvious mechanism that would directly suggest that is that the light comes directly from blackbody luminous emitters of various temperatures.
But in fact, the source of all that light is one emitter (ol' Sol) and the reason we see different chromaticities at different angles of observation, or at different times of the day (different solar elevations), is a result of the "filter" effect of atmospheric phenomena, I think mainly scattering (if we discount smog and such).
I can't imagine any simple model of this that would suggest a chromaticity of the observed (filtered) light that would track the Planckian locus. (Perhaps you have some mechanism in mind you did not mention.)
As to the matter of much of the departure being in the purple direction, we are well aware that the scattering often moves the chromaticity of the light in that direction - the reason that, as we contemplate "daylight" from a cloudy situation, its CCT is higher than that of open sky daylight.
Just some thoughts.
I should clarify, the high corellation is to the line of best fit gamut passing through my data.
The corellation deteorates towards purple and red. I think scattering also.
I can't think of anything that matches the Planckian locus either (which was the next question
) - I know of the Daylight locus and another one found for the skies over Granada - mine is similar, but given the difference in conditions Tokyo has over other places - I am not entirely surprised that it is different.
It never occurred to me that consideration is given to cloudy skies for that reason - even though it has been staring me in the face for a while - that happens to me all time - lol.
So, a summary of 3 major causes of 'purple' shifting could be summarised as:
1. White balance
2. Aerosol
3. Twilight
Am I correct here?
Once again, thank you very kindly for the replies, they have confirmed a number of notions I had and helped clarify a few more. I very much appreciate this.