Star photography - trails or no trails
Sections in this page
In the days of film there was no possibility of stacking multiple exposures. Astrophotography required very long single exposures. That meant a motorised mount was essential, to compensate for the Earth's rotation and stop the stars trailing. Exposures for an hour or so meant that a guiding system was important too, because no mount is accurate enough over such a long time.
With an ordinary digital camera plus some free software (such as my own GRIP) a completely different way of working is possible. No guiding system is needed. Digital cameras are so much more sensitive than film that very short exposures can be used, but many of them are stacked together. That makes it possible to start with a fixed, unmotorised tripod. Anyone can now take photos that a few years ago would have required very expensive equipment.
These pages aim to explain how.
There is a more recent and more detailed set of pages starting here (a talk I have given to various clubs).
Star trails
If you put your camera on a tripod on a dark clear night, set it to B (the "Bulb" setting) and open the shutter for several minutes you may get something like this:
Canon EOS5D 24-105mm lens @ 24mm 641s f/4 ISO200 2007:2:11 18:36:57
That is a wide-angle 10½-minute exposure of the constellations of Auriga and Gemini, with part of Orion below and Taurus on the right. The stars form trails because of the Earth's rotation and the fact that the camera is standing stationary on it. The Earth rotates by 15 degrees per hour so these trails are about 2½ degrees long.
Notice that the stars quite clearly have different colours. The human eye cannot see that so easily in the dark. Then notice that the colours are green and blue without much hint of red. That is sadly because of light pollution. The original image had a background of reddish fog due to suburban lighting which we have removed by processing with GRIP. Here is a scaled-down version, to fit in this page, of the original:
Canon EOS5D 24-105mm lens @ 24mm 641s f/4 ISO200 2007:02:11 18:36:57
Now you can see something else: the background is not uniform. This is partly due to a vignetting effect of the camera lens, perfectly normal but usually less obvious in daylight scenes, and partly because the street lighting does not produce a uniform reflection from dust in the atmosphere. GRIP can do background correction automatically and that was done to obtain the first picture above.
Here is a full-scale portion of a longer exposure:
Canon EOS5D 24-105mm lens @ 24mm 1817s f/8 ISO200 2007:02:11 18:47:55
The bright star here is the uppermost one in Auriga (see previous photo). This image shows another problem: camera noise. This is perfectly normal and occurs in all electronic devices. It is due to random thermal movements of electrons. The fact that it is thermal means that it gets worse at higher temperatures. Some types of astronomical cameras, based on CCD chips, can be cooled in liquid nitrogen to reduce the noise. Consumer cameras are not designed for that.
The good news is that thermal noise is random. So the speckled pattern we see here will be quite different from one exposure to the next. To get rid of noise we can therefore average several successive images, something which GRIP is designed to facilitate - see the batch menu or the table of images. However, for the kinds of photos we have considered so far the subject will have moved on. We need to move the camera to follow the stars rather than fixing it in relation to the Earth.
Avoiding trails
Before we look at moving the camera, consider that another approach is to take a much shorter exposure but increase the sensitivity of the camera and use the largest possible lens aperture. The next photo was taken in just 10 seconds but at f/1.8 and ISO1600. A fixed lens was used rather than a zoom in the hope of collecting more light. Zoom lenses generally have many more elements, so they contain more inter-component surfaces with the potential to reflect or scatter light. The author has not yet done any quantitative comparison of zoom and fixed lenses in this respect but he is hoping that GRIP will enable someone to do so and conclusions may be reported here.
Canon EOS5D 50mm lens 10s f/1.8 ISO1600 2006:12:06 17:54:11
At first sight this photo of the Pleiades and surrounding areas has nice sharp stars, but when we look at a portion of it at full scale* some elongation of the star images due to the Earth's rotation becomes apparent. This would be worse at longer focal lengths:
Canon EOS5D 50mm lens 10s f/1.8 ISO1600 2006:12:06 17:54:11
So it depends what you want to do with the photo as to whether a short exposure at high sensitivity and wide aperture will suffice. This photo was also a lucky shot because there happened to be a meteor or perhaps an artificial satellite in the frame.
In subsequent pages we will look at how to make a longer exposure without trails.
* By "full scale" I mean zooming in on the image so that 1 pixel in the camera equals 1 pixel on the screen here. Generally that also involves cropping the image to fit on the page, so we see just a portion of the original.
Star trail length calculator
This assumes your camera (with lens or telescope) is in a fixed position.
The image size is initialised in the last section of the form to be 1024 pixels by two thirds of that because that is the kind of size that would be seen on a computer screen. Change the size to that of your camera sensor or of a print if you want to see how big the trails would be in those cases.
Example: in the last photo shown above (the Pleiades) the focal length was 50mm, the camera was full 35mm frame, the exposure time was 10 seconds and the declination* of the Pleiades is about 24°. Entering those details into the calculator we get these results:
Field of view = 39.6 x 27.0 degrees [the whole uncropped image]
Trail length = 0.0382 degrees
= 0.000964 of field width
= 0.00141 of field height
If we leave the image size at 1024 x 683 pixels, the final line is:
Trail length = 1.45 pixels
So that is why in the first version of the image above, scaled to fit the screen, the stars look like points. But if we zoom up so that 1 pixel in the camera equals 1 pixel on the screen but crop to fit the image on the screen, as was done for the second version above, the image size has to be the detector size of the camera (4368 x 2912) and then the last result from the calculator is:
Trail length = 7.59 pixels
And that is of course noticeable.
You can use the calculator to find out how short an exposure is needed to avoid significant trails in your circumstances.
Don't forget that shortening exposure time means that either aperture or ISO sensitivity, or both, must be increased in order to capture the same amount of light (as explained on my camera techniques page). Increasing aperture means focussing must be more accurate but that is difficult at night. Increasing ISO sensitivity results in more noisy images.
So we need to look at how we can do better, even without moving the camera to follow the stars.
* Declination is celestial latitude, the angle above the celestial equator. You can find that from a star atlas or, for many objects, by searching on the Internet. However, if you don't know the declination of the centre of your photograph you can just leave 0.0 as the value in the calculator. In our Pleiades example that would have made the final result 6.76 instead of 7.59 pixels. Declination becomes more significant for areas near the poles, where trails for a given exposure time are much shorter.