Parsing Picasso’s Pastels

Oil pastels are a surprisingly recent ‘traditional’ art medium.

Developed independently by Sakura (Cray-Pas) in Japan in 1927 and Sennelier in Paris in 1949, they remain far less popular than dry pastels. Although Sennelier relishes telling you that it was Picasso who asked them to develop and supply oil pastels, to the best of anyone’s knowledge he has left no works painted predominantly in this medium.

Being fairly novel, there are considerable differences between oil pastels supplied by different manufacturers; even their consistency varies from Sennelier’s soft and buttery lipstick-like sticks to Holbein’s harder bars.

Colours are still more contentious: Holbein leads the field with no less than 225, but these are actually 45 colours each in five different intensities, whilst Sennelier offers 120, including a huge range of greys (apparently Picasso’s wish) and iridescent colours. As you might expect, whenever two or more oil pastellists are gathered together, there are three or more different opinions on the relative merits of the different brands, including their opacity and hues.

Quantifying and comparing colour is actually much tougher than nailing jelly to the wall.

Purists – and traditional art thrives on obsessive purism – might seek the darkest dark, only to insist that they will never use anything that is actually black. Should they look for a Caran d’Ache Neopastel, or one of Sennelier’s pedigree greys?

Having complete sets of each, I thought that it should be simple to provide an objective answer. I was already making myself complete sets of colour patches, so I scanned them into my Mac and got Adobe Photoshop to show me RGB histograms for each of the darks.

It was then that everything became bogged down.

I had not built a colour profile for my scanner, so scanned in an IT8.7/2 standard test chart to check that it would provide reasonable comparisons even if not entirely linear. That led me to question how good my scanner was, so I went out in search of something better, like an Epson Perfection V700 or 750.

Walk-in retailers seem to have decided that no-one now buys separate scanners or even printers, and what I wanted was an all-in-one – which was exactly what I was trying to upgrade from. Even the pressures of the recent recession seem unable to convince retail chains that customers like to get what they want, not what the retailer wants them to have, so I purchased online.

Whilst mulling over how best to build a profile for my new scanner, I stumbled across the ColorMunki Design, a zippy little device that claimed to be able to calibrate my screen (already Hueyed), printer, projector, and tell me the colour of any surface or sample patch. So no sooner had I scanned in my colour patches at some ludicrously high dpi and in 48-bit colour, than I was measuring them directly.

The ColorMunki quickly impressed: plop, click, and another patch was added to my growing swatch. I then read RGB and L*a*b* values for each patch, and recorded them in a Numbers spreadsheet. For all its immaturities, Numbers shone, as it let me paste in a swatch of each oil pastel from the scan, something completely lacking in Excel.

Being a professional sceptic, I also tested the ColorMunki against my brand new 5” x 7” IT8.7/2 standard colour target. Differences between the measured and claimed colour values on this target were within +/- 1 unit for L and a channels, and +/- 2 units for the b channel: not bad considering all the inherent problems with targets and colour.

You can download the end results in PDF format from this article.

I am still not sure whether the jelly is firmly nailed to the wall, but the next time that someone wants to know what colour Sennelier’s Bordeaux (number 008) oil pastel really is, the answer is there in L*a*b*.

Updated from the original, which was first published in MacUser volume 25 issue 05, 2009. Since then the ColorMunki Design has proved invaluable, although I dearly wish that X-Rite would maintain its software better.