Hi Everyone
I have meant to chime in on the discussion, and Dave's comment seems most apposite. There has been some very interesting ideas put forward and I hope my thoughts add something here.
The images were most likely to have been taken on Kodachrome second version which was released in 1936. I think it is highly unlikely he was using Agfacolor (the only other contender) in British territory due to the war. I have been unable to find any information on the spectral sensitivity of this version of Kodachrome and can only assume it was similar to the confusingly named Kodachrome II introduced in 1961. As Dav points out Kodachrome II when scanned is considered to have a blue cast.
As I see it we have seven impacts on the 'colour' of the image:
1. - the 'absolute' colours of the scene
2. - the ambient light at the time of recording
3. - the spectral sensitivity of the recording device (i.e. lens & filters)
4. - the spectral sensitivity of the recording medium (i.e. film)
5. - the 'fixing' process (ie.e development)
6. - the reproduction process (scanning, printing etc)
7. - the viewing environment (monitor calibration)
We are attempting to unwind this knot backwards. Fortunately, it is not as hard as it sounds to get a bit closer to what the absolute values probably were.
First, we can establish the ambient light at time of recording. We know Singapore is on the equator so there is no seasonal variation in light due to the angle of the sun through the atmosphere, unlike Northern Europe where there is a big difference between summer and winter. From the shadows we can tell it was photographed in the mid-afternoon or mid-morning. The only unknown is the amount of moisture vapour in the air. I live in Singapore, so I chose a day with similar cloud cover to measure the ambient light. Interestingly it was 5500 Kelvin (compared to 6500 Kelvin in Europe).
Secondly, we can only hypothesis about the influences of the devices & media (2-6). We can take a stab at this if we know the absolute values of the scene. Obviously, as this is what we are trying to work out the argument is a bit circular. However, my first reaction on seeing the images on a calibrated monitor (i.e. correcting for item 7) was that grass in Singapore isn't that colour.
To make an estimate of the colour shift, I compared the grass at Sembawang to a set of Pantone Colour swatches. These are an absolute value, and the grass hasn't changed here, so there is a good chance the match will be useful. The shadows were a good match to 574C, midtones to 575C and highlights to 576C. I now realize I didn't update the swatch labels on the images, but they run 574C, 575C, 576C L to R.
I then made some adjustments to the photos. First, increasing the exposure by about 1 stop, next correcting the white balance (from a known neutral - the white on the roundel), finally adjusting manually to match the hue of the grass with the Pantone colours. I did this by choosing zooming in on one area of shadows & highlights, then sampling its colour before and after adjustment. I think the result is not perfect, but somewhat better, particularly the shadow values.
I leave it to others to draw conclusions about the colours compared to various standards and paint shades, but my thoughts are:
- to keep things consistent, measuring the paint chips and samples using a 5500K light is appropriate. Most colorimeters and spectrophotometers default to 6500 K [D65] or 6000 K [D60]. The measured chip can then be very simply compared to the image in a color managed environment such as Adobe Photoshop. I personally would love to see this, and am happy to send the original TIFF files I made to anyone who wants to play with this.
- some paints and coatings are near impossible to measure with colorimeters or spectrophotometers. I was fortunate enough to be shown a piece of Catalina skin by Ian Baker which if viewed from one angle was blue and from another had a greenish tinge.
At the end of the day, unless you are there with the colour measuring equipment we cannot get total accuracy. We can just try to make our margin of error smaller.
Hope this is helpful
Christopher
Original Image
White Balance & Exposure Adjusted
Tint Adjusted