What is Dithering?
In astrophotography, to dither means to shift the pointing direction of the telescope/camera, only slightly, in random directions between exposures. This action allows hot and cold pixels, cosmic ray artifacts, fixed pattern noise, and satellite or airplane trails to be removed during the stacking process.
This is because dithering displaces the image sensor in regards to your target. Hot pixels, which are individual pixels that look brighter than they should lie in the exact same place in every exposure.
By moving the pointing of the telescope, dithering shifts the stars to a slightly different place in each frame. As you process your images, you will align and stack individual frames based on the stars in each image. This shift places the hot pixels in a different place within every frame.
This subtle, and easy to initiate feature found in most camera control software applications, can make a dramatic difference to your final image.
The Bubble Nebula. Dithering took place between each exposure, using PHD2 Guiding.
The Sigma rejection stacking method, available in most astronomical image-processing programs, applies a mathematical algorithm to examine every pixel and discard outliers that are very different from the average in your group of images.
This means that if you have 10 frames and a hot pixel is only showing up in one particular location on a given frame, the algorithm replaces it with an average from the other frames. For the Sigma algorithm to work, you need a minimum of 10 frames.
To get a better understanding of what dithering does, imagine that you have a single star captured in a specific position on the sensor.
Autoguiding should make sure that this star is in the same location for every shot. But what if you throw in a hot pixel in another position that remains there in every frame? When you dither between image frames, the star will not be captured by the same pixel, but instead the pixel at in a slightly different location on your sensor.
The hot pixel, however, remains in the exact same position each time. When you stack the frames (using software such as DeepSkyStacker), the registration process aligns all of the frames. This will align the stars together in each frame, which were captured by slightly different pixel positions each time. The hot pixel will have shifted as well because the stacking software registers the image according to the “moving” star positions.
This means that the stacking process can easily identify hot pixels and get rid of them using a pixel rejection method.
Using Astro Photography Tool and PHD2 Guiding to dither between each image exposure.
How to Dither Your Images
In the days of film, one may have used an SBIG ST-4 autoguider, and manually triggered a cable release for every exposure (we are spoiled these days)! You would dither by turning off the autoguider, then manually move the scope a little with the hand controller, turn on autoguiding again, and then start another exposure. This, however, would be scrupulously done throughout the entire night.
I experienced this on a small scale when capturing images of the Carina Nebula from the resort I was staying at in Costa Rica. I was using short exposures, an aggressive ISO setting, and I needed to do something to help reduce the noise in my final image.
I manually dithered this image of the Carina Nebula using an iOptron SkyGuider Pro mount.
Thankfully, today, we are able to do automate the process of dithering astrophotography images digitally. We now have the luxury of using a computer to control the telescope mount, primary imaging camera, and autoguider.
The software that controls the different pieces of hardware enables everything to communicate with each other. Programs like Astro Photography Tool, BackyardEOS, or BackyardNIKON can work with PHD2 for autoguiding. You can also use Sequence Generator Pro with either PHD2 or MetaGuide. High-end software including MaximDL integrates all these functions, although I have not tried this one myself.
The camera-control software pauses the imaging sequence between each exposure, and sends a dither offset to the autoguider program. This then passes the command on to the mount control driver, moving the position of the scope. The autoguider software re-acquires the guide star to begin guiding. Then the camera-control program will start taking the next picture.
The aggressiveness setting controls how far the mount pointing will move between exposures. You may want to increase this value a bit more if you are using a DSLR or one-shot color CCD camera to compensate for the 4-pixel Bayer color filter array overlaid on the detector.
The same idea applies if the optic you’re guiding with has a longer focal-length than your imaging scope, which might happen if you’re shooting with a telephoto camera lens on top of your normal telescope. Without a big enough shift in pointing, your pixel offset will be too small to make a difference when stacking the results.
If you aren’t guiding, and in some cases even if you are, you may notice your stars moving between frames from polar misalignment or flexure between your guide scope and imaging scope. You might think you can use this natural drift for dithering.
However, this generally will not work. Dithering needs to be done in random directions each time to be most effective. Any drift from polar misalignment or flexure is usually in one direction, therefore one would end up with smeared pattern noise (walking noise) appearing like dark and light streaks in your stacked result.
Benefits of Dithering
There is a saying by professional astronomers; “dither or die.” It may not be quite that serious for amateur astrophotography, but dithering will improve your images tremendously. Have a look at what my good friend Dylan has to say about dithering:
Dithering increases the all-important signal-to-noise ratio and removes artifacts like hot pixels and satellite trails. Dithering can even compensate for temperature mismatches between darks and lights shot with an uncooled DSLR.
Every ‘fixed pattern noise’ will benefit from dithering, in particular, the thermal noise (dark current noise) we have to battle using uncooled DSLR cameras. This thermal noise is of a fixed pattern, and thus shifting the image a bit over the pixels of the sensor can greatly benefit you because you will be able to get rid of this thermal noise in the stacking process.
Apart from improving the noise levels of your image, one will also benefit from the fact that when you shift around the image over pixels you will see that they may not all perform equally well. You will notice some pixels will be dead, cold or hot and therefore the signal there won’t be picked up as good as on other pixels. Additionally, some pixels are more sensitive than others.
If you didn’t image, those signals that end up on those less efficient pixels would not be picked up as clearly as on the other pixels. Shifting this around will greatly benefit your final image.