I am exploring a new processing technique that started out to be about getting better stars in my DSO images. As I thought more deeply about it, I realized that there is also an opportunity to get better target objects in my DSO images as well. For this reason, the objective has morphed into being better DSO images. While there are other opportunities to improve DSO images, the present technique that I am pursuing is based on separate handling of an image’s star field and the target object, and then combining the fully process stars and target object images at the end.
My traditional exposure durations for DSO images in a light polluted sky have been about 60s, which seemed to be a good balance between getting the skyfog peak above the noise while minimizing clipping of bright star centers. At first I saw suboptimizing (by clipping) star data as a necessity for getting good DSO target object data. And, I was pretty happy with my slightly blown out stars in a recent Monkey Head image and last year’s Veil Nebula images. I was also pretty happy with the target objects in those images as well.
Recently knowing that tools exist that permit me to remove stars from a DSO image, I read up to see what could be done to improve star data in processing. What I learned surprised me. There is an even bigger opportunity to improve star data by starting earlier in the value chain and collecting data for the stars component of an image separately from the longer capture for the DSO image. Essentially, 20s light frames is all that is necessary to get the an image skyfog peak out of the noise, with only the very brightes stars being clipped. I was also suprised to learn that only about 30 to 60 minutes of integration time is necessary to produce a good quality star image. After completing the separate processing for star data and target DSO data, the two images are combined.
Here’s the epiphany that fell on me: With separate capture of star data and DSO target data, there is no longer any need to foreshorten my main DSO capture to just 60s for the sake of stars, which I intend to remove from the DSO component image anyway. With longer exposures, I can accumulate weak DSO signal at a faster rate than with shorter exposures. For example, given a choice between an hour’s worth of 60s exposures (1 hour integration time), and an hour’s worth of 120s exposures (again, 1 hour integration time), 120s is a better choice because my data will have a better SNR in the longer exposure.
I don’t know what the longer exposure duration number will be until I experiment, but in principle, I can get better DSO data by exposing longer, and I can expose longer because I am not worried about the star component of the long-duration images.
At this point I have captured short exposures, removed the nebula, and processed it as stars only. The star color is great. I think that it can be better if I add luminance. The background seems too smooth. I want to play with processing parameters that might allow a little noise to make the background more realistic.
The next step in my exploration is to capture and process long duration images for the DSO component of a final image. I am thinking something on the order of three or four minute light frames. Experimentation is required. Click image below to see in Flickr.