Ordinary Meeting, 2005 January 26


Wide-Field Astrophotography

Mr James recalled that in the past he had been something of a sceptic regarding digital cameras: they had seemed relatively insensitive, and to have rather small fields. He was one of many who had chosen instead to dedicate his attention to following the development of new types of film, each supposedly giving superior performance to its predecessors for long exposures. However, he proceeded to show an image which had caused him to think again: a wide-angle image of the Milky Way, which Nigel Evans had taken during their expedition to Sharm-el-Sheikh in Egypt in 2004 June to view the transit of Venus. The quality of this image, taken with an 8mm aperture lens and digital SLR camera, had convinced him that technology had moved on, and on the grounds of it he had since bought his own camera. The speaker remarked that he had often heard the question asked whether "film was dead", and in contrast with his previous opinions, he was now convinced that digital technology had reached a stage where it was challenging even the final remaining strengths of film.

The speaker's own camera was a Canon 10D, already an obsolete model, though he noted it had a consumer counterpart, the Canon 300D, which was essentially the same, except that it was packaged in a more compact and somewhat less sturdy case. Essential for astrophotography was a very high-quality lens, as it was relatively difficult for them to focus stars into point sources without any appreciable spreading of the light. Mr James recommended the eBay® website as a good source, remarking that he had himself found a very good lens there, priced at £160. The trend seemed to be towards increasing popular demand for zoom lenses, and so the fixed-focus variety could often be found quite cheaply.

The CMOS sensor array in the 10D was found to have quite a low noise-level, and without any cooling, exposures of up to five minutes were possible before it began to become an issue. The chip itself contained 7.4μm-square pixels in a 3,072 by 2,048 array, measuring 15.1mm by 22.7mm in total.

Mr James remarked that whereas CCD-users needed to carry a laptop with them to drive their cameras, digital camera users had no such need, as the body of the camera itself read and stored the data from the sensor, whilst at the same time being somewhat less bulky to carry around. The capacity of consumer flash memory cards was growing incredibly fast, so that a card with plentiful storage for a night's observing was now available at minimal cost. The control of the speaker's camera was very similar to that of a standard SLR – for example it had a 'bulb' setting for long exposures. By default the camera returned images in JPEG format, though it also supported the return of raw data directly from the sensor. The speaker reminded members that the latter option was essential for astrophotography, as it allowed tasks such as flat-fielding and the removal of hot pixels to be undertaken on the images. By comparison, these were impossible after the image had been compressed into JPEG format, as information from each of the individual detector elements were blurred together.

The speaker had found the LCD display on the back of his camera to be remarkably useful as a focussing-aid since it could display the image at a very high magnification. Whereas older SLR cameras had invariably provided a split-prism in the viewfinder to aid with reaching sharp focus, their modern counterparts tended to lack such niceties on the assumption that automatic focussing was a ubiquity. In addition, it appeared that the focussing wheels of all Canon lenses could now be turned beyond infinity, which meant that it was no longer possible to simply turn them to the end-stop when doing astrophotography.

Mr James commented that the 10D, along with many similar cameras, had an ISO setting, allowing the user to make the sensor simulate the behaviour of film of a given speed. He remarked that for astrophotography one might intuitively choose to use a high-sensitivity setting. However, he explained that this was actually not a good choice. When images were read in raw format, each pixel directly returned a 12-bit number (4096 quantisation levels), giving the number of quantised units of brightness detected. The ISO setting increased the gain between the CMOS sensor and the sampler, but beyond a certain setting there was no benefit, since the image noise was already covering several quantisation levels. In fact higher ISO merely had the effect of reducing the dynamic range available and restricting the maximum exposure time in light polluted environments. Thus, best results were obtained by configuring the camera to a relatively low ISO setting (200), and taking a longer exposure.

The 10D supported the use of a programmable digital cable release, which made it straightforward to schedule the exposure of a series of images at given intervals. The speaker noted, however, that photographers who were happy with old-fashioned push button releases would have to pay £30 for the digital equivalent. This seemed an extortionate price, though Mr Maurice Gavin had reportedly soldered together such a device himself, with minimal cost or effort involved.

Moving onto software, the speaker recommended the use of a package called Iris, which was freely available online.3 For new-comers to the field of image processing, there was an extensive online tutorial at the same website. The software itself supported the input of images in the raw data formats used by both Canon and Nikon cameras. As an example, Mr James demonstrated what he had been able to achieve with a single 180s exposure of Comet Machholz. The raw image from the camera suffered a significant transverse gradient as a result of shadowing caused by the reflex mirror mechanism, as well as a few blotches caused by dust on the sensor. To counter these blemishes, he had taken an image of the flattest field he could find, by directing the camera at a blank region of the sky in twilight. The comet image was then divided by this, and any residual background variation was identifed by letting Iris fit a polynomial to the result. The calibrated image was then divided by this smoothed flat, and both the dust and ion tails were then clearly visible, despite their brightnesses being a mere ~1% of that of the sky background. That the 10D was able to obtain such detail from a sub-optimal suburban location seemed an impressive feat.

To conclude, the speaker mentioned one downside of the 10D. Although CMOS image sensors were typically quite sensitive well into the infrared part of the spectrum, the 10D incorporated a filter in front of the sensor, which blocked much of this light. The reason for having such a filter was that the focal length of the optics differed with wavelength, and so if infrared light were detected, it would not be in focus when visible light was, causing a blurring of the image. However, the filter's cutoff was around 20nm shortwards of the Hα line at 656nm, making it a poor tool for imaging hydrogen emission. To give an example, the Horsehead Nebula in Orion would be virtually invisible. However, the speaker concluded that on balance the Canon 10D was an outstanding imaging tool for £700. Given how unimaginable its imaging capabilities would have been only a very few years previously, he wondered what technology might become available to amateur imagers in years to come.

Following applause, a member asked Mr James how easy the sensor was to clean. He replied that Canon's official advice was that it should be returned to them for cleaning, but that as this seemed somewhat impractical, he had personally chosen to use a blower brush to blow air over it, without any physical contact. This procedure had seemed adequate so far. Whilst researching what others had done, he had found various accounts on the internet, describing the passing of various cleaning fluids over the sensor, though he did not envisage he would be risking this himself. He also remarked that he had read similar accounts of people dismantling the camera to remove the infrared filter – though this idea also failed to entice him.

The President then proceeded to introduce the evening's final speaker, Dr Nick Lomb, former Vice-President of the Association's New South Wales Branch, and current Curator of Astronomy at the Sydney Observatory.






Color scheme