AAO image reference 'calibration'.
Image and text © 1977-1995 Australian Astronomical Observatory, photograph by David Malin.
Click 1024x678 for a 768 x 1024 version of this image, 1080x1990 for the full HD version.
Also available as a Powerpoint or Keynote file
If your monitor (or your data projector) shows all 23 tone gradations in the vertical scales and the scales are neutral from end to end, you will see the astronomical pictures on these pages as they were intended to be seen when the RGB process that made them was developed in the mid-1970s. If you want to know more, read on.
All the astronomical images on this site were made using an additive 3-colour process derived ultimately from James Clerk Maxwell's 1861 demonstration of the possibility of colour photography. I developed the version used for these pictures at the Australian Astronomical Observatory, beginning in 1977. At that time there was little information about how to set up an RGB system from scratch, so I devised my own.
Since I intended to make my astronomical pictures from three monochrome (B&W) negatives exposed through standard astronomy filters to capture essentially blue, green and red (RGB) light, the passbands I would have to deal with were well defined. However, there was no calibration source in the sky to show convincingly that my three-colour process was revealing anything like true colour. The astronomical objects themselves are too faint to be seen in colour, even with a large telescope. To check and calibrate the method, I made test exposures in the lab on panchromatic B&W film, using filters to replicate as closely as possible the astronomical passbands. The subject matter was a set of Kodak colour patches and a greyscale, then widely used as colour references.
For the photometric calibration of plates, and completely unconnected to the problem of colour calibration, built into the AAT prime focus camera was another reference greyscale whose image was projected at the edge of the plate. I adapted this so that the light source was equivalent to mean noon sunlight (5500K), the same as the zero point I had chosen for my additive colour system. It is normally off the image area, but on extremely wide field images (e.g. AAT 38a and AAT 103) it remains visible. If this greyscale is neutral in the final 3-colour images the system is internally consistent and has an external reference point.
The (6 x 9cm) B&W greyscale negatives made in the lab were contact copied to produce sets of matching B&W positives, each set with a different contrast. In the darkroom I combined them in register using the standard RGB Wratten filters (initially using Cibachrome) to obtain good colour balance, with all the steps of the greyscale visible (and neutral), and with colours that matched the original subject quite well. This is quite a complex business, since the additive process allows full control over hue, contrast, saturation and colour temperature zero point in a way that normal colour photography (thankfully) does not.
Later (after 1985), 3-colour images were made by projecting B&W positives on to 8 x 10-inch colour negative film, and again the greyscale pictures were used to adjust darkroom processes to achieve consistent results. A decade later, when I began to digitally re-master the astronomical images from the original data, I again used the same refrence negatives. With the digital era, RGB colour image-making was suddenly in fashion. In the image above my original 1977 reference images has been improved with (vertical) greyscales of finer gradation and a horizontal spectrum of RGB colours added digitally. The spectrum is only realisable on a colour monitor (or a good data projector), however, the colour patches and greyscale are from the original 1977 negatives.
Elsewhere on these pages there is general discussion on how photography works as well as a brief technical overview of other aspects of the three-colour process. There is also a a pointer to early references on this topic. The image above is very useful for setting up digital projectors so they reproduce photographs (as opposed to data) acceptably. Click here for a 768 x 1024 version of this image and here for the image embedded in a Powerpoint file.
David Malin, 2020, January 21