The production of colours has a curious history, one that intersects the histories of art and science. Imagine painting the Thar desert with its vast rolling mounds of striated sand dunes. If you want to recreate the sand dunes on paper, the colours depicted accurately, you could pick up a fistful of sand, grind it into a fine powder and bind it with water or an oil-soluble medium. The method allows you to render a colour very close to the original on paper or oil. Yet this small success immediately poses a conceptual problem. To understand that, you must imagine a sand dune as a three-dimensional object with the sun shining down upon it. Some parts of the dune will be brightly lit, others less so and the rest covered under shadows. Will all the three parts have the same colour? If you observe any evenly coloured object around the house which is under both light and shadow, you will realise that a large part of the colour we actually see emerges out of the interaction of light and the actual pigment in question. Besides, people have varying sensitivities to colours, the most obvious example being that of colour blindness in a significant section of the population.

Both scientists and artists have raised key questions about the nature of colour and their concerns are finding increasingly common ground. But first let us return to painting a scene made up of a uniformly coloured object in varying light. The sand dune cannot be rendered in accurate colour by just grinding sand particles to make a pigment. Clearly, we need other colours. Artists have banked on a large palette of colours and mixed them up to make different shades, hues and tones. In a realistic painting, the colours for a particular scene are identified by the eye. Here it is experience and intuition that guide the mixing of colours. Now that is a very hard task, one that takes effort to master. Once the landscape of paints has been explored by painters, they can take multiple routes to get to their art — from the hyper-realistic to the surreal. Many great works are a flight of imagination — accurate colours be damned. But there is one area where even the most creative artist will struggle with the accurate rendition of colours. If you are an art restorer tasked with the restoration of a Leonardo da Vinci painting, you will be required to match colours, and pretty much everything else, as closely as possible to the original.

This is a conundrum, especially given that memory is fallible. So materials and their colours must be documented photographically at various stages of the restoration work. To understand the complexity of the exercise, take 10 different cameras and shoot a picture of an object or a scene under the exact same lighting. Observe them in the computer monitor one after the other. All 10 images will look different, some significantly so, in the rendition of their colours. Now if you were to pick a single image from this set and see this on different computer monitors, you would again see that the image looks slightly different on each monitor. Like humans, cameras, monitors and even colour films ‘see’ and render the same scene differently. Digital imaging, in fact, is not the panacea that some imagine it to be. Yet the digital colour camera technology has many advantages and allows large-scale accurate reproduction of colour as well as massive distortions of colour in a way that was not possible previously.

But before we move on to accuracy in colour, we must take a step back and realise that colour films — Kodak, Fuji, Agfa and Konica — saw colour in conceptually the same terms as artists saw and continue to see it. The concept, briefly put, is this. Any colour can be made by mixing varying quantities of three colours — red, green and blue. While artists made this mix typically on their palettes, colour films worked by having three different colour sensitive layers one on top of the other. Light would expose these layers, one sensitive to red, the other to green and the last to blue, to produce three images stacked on top of each other, which could be combined to produce a full colour image.

The pixels in most modern digital cameras are not three overlaid red, green and blue pixels. Instead each pixel can only record the intensity of only one of the three colours. The single colour pixels cover the entire sensor, typically every row repeating a pattern of individually coloured dots such as RGGBRGGBRGGB... What results is a speckled grayscale black and white image from which the full colour image must be guessed, pixel by pixel. Colour is now calculated using mathematical algorithms which use a knowledge of how the world actually looks like with some moderately sophisticated mathematics. The fitting of so many colour filters over a small sensor surface is a tour de force of modern technology and the algorithms that compute the image have revolutionised imaging. All of this is a feat of mathematics and computer science combined which when proposed by Bryce Bayer, a scientist with Kodak, over 40 years ago, was considered too far fetched. A few decades later, digital colours were inferred by algorithms that in their guessing of colours are a throwback to the mind of an artist more than that of a physicist.

BLINKSANTANU

Santanu Chakraborty

 

Santanu Chakraborty is a Bengaluru-based engineer, scientist and photographer