, , , , , , ,

I’ve been playing with hair today. Hair fascinates me: we compare it to feathers and it seems downright dowdy but, when we draw hair as dull as we imagine it and then compare it to real hair, our rendition is usually more drab than hair can possibly be.

What I did for this test was to take a photo of a very standard male brunette hominid in decent but uneventful lighting* and cut out his hair. I pasted it on a pure black background and passed it through some code I prepared earlier.

Diagram a: the brunette hair cut-out.

The code read the file contents and processed all the colour data, plotting two graphs: the one on the left is hue (horizontal plane) against brightness (vertical); the one on the right shows saturation (horizontal) against brightness (again vertical).

Diagram b: the scatter graph


If you look at the scatter graph on the right, you can clearly see a very limited range of colour saturation. The image in its entirety stays very grey but it’s more colourful at the darkest range and most colourful at the brightest range. Mid tones all stay very muted.

In the scatter graph on the left, you can see that all the colour action takes place in the orange and yellow end of the spectrum as you would expect. Nevertheless, blues, greens, and violets are present at every brightness range. There is a great deal more colour in brunette hair than we see at a casual glance. Part of this fine scattering of unanticipated colours can be down to the photograph’s original jpeg format, which is considered a ‘lossy’ image format because it dithers colour somewhat to reduce file size. The quality of the saved file was set to high though, to minimise loss. The other cause for these colours is that hair, by its nature, is glossy. As a result, surrounding objects are subtly reflected in it. I think what we’re seeing in the blues and greens is a very minute reflection of the grass and sky – by far the dominant features around our hominid.

I shall try this test again once I have a digital camera that can save directly to a lossless png format, and I want to try it again with some more exciting hair. I’m keen to see how ginger hair performs under analysis. I suspect a plotted hue brightness chart that leans from orange to yellow the brighter it gets.

Applications of this technique: I can use it to create a gamut map of available colours for painting a subject; I can use it to compare the same subject under varying lighting conditions (given a digital camera that doesn’t colour adjust).

* By this I mean that the photograph was taken outdoors, midday while overcast – avoiding a huge gradient of undesirable blue reflections. No lightning, fires, aurora borealis, or artificial lighting of any kind.