Hi Bud,
Now, I hope nobody takes this as any thing more than a simple discussion of ideas/methods in processing.
Naturally. One of the main goals of this forum is to promote the development of astrophotography through constructive criticism and open discussion. In general, you'll see few attaboys here. You'll see many vivid discussions on key topics, including technical, aesthetic and philosophical discussions on image processing, processing ethics, photographic art and software development. We try to avoid all taboos here.
I would say that the same criterion would apply as a person that uses a star mask....deconvolve/HDR transform and see what shows
We pursue a systematic approach to image processing. Before applying deconvolution, for example, we first analyze the data to see whether we have enough signal. We frequently perform wavelet decompositions to inspect or analyze the distribution of noise in the image at different scales. Ultimately, our accumulated experience tells us what path to follow. We also love planing and trying out new paths. In general, we try to be very careful to avoid enhancement of marginal data because that leads to generation of false structures or artifacts. If we don't have enough signal to apply deconvolution, we seek for other algorithms that can be less sensitive to noise, such as wavelet transforms. With HDRWaveletTransform we basically try to overcome dynamic range problems. This is a key topic in astrophotography because we often work with data sets that represent structures with numerical values that differ by many orders of magnitude.
In general, most of these algorithms usually require masks to protect low SNR data. You can deconvolve the core of a galaxy, but you probably can't deconvolve the structures on its outer spiral arms in the same way. Selective processing is necessary because the uncertainty in the representation of object structures is not constant throughout the whole image: it is a (inverse) function of incident light. And here comes our eternal companion —the noise, what else— to make the things funny. There are other instances where the nature of the data acquisition processes and the instrumental limitations reduce our possibilities to process some image structures. This is the case of the stars, for example, where we frequently need special masks to avoid generation of artifacts due to jump discontinuities (ringing), or to avoid flattening of star profiles (saturation).
Of course, you can always make the decision to not deconvolve, not sharpen, not perform noise reduction, not try to fix dynamic range problems, not handle image structures at different scales, and apply just basic nonlinear intensity transformations to adapt the data to the human vision system. You can do that, but then you should ask yourself why you are doing astrophotography in the first place. Without a personal interpretation of the data, without pursuing the maximization of information representation, there's no fun at all. When we don't take our responsibility, everything is secure but also very boring. When we face the problems, we take the risk to make mistakes, but everything is exciting and we grow. Astrophotography isn't too different from everyday life.
So once we can agree in that we need image processing in astrophotography, and in that we want to squeeze the last bit of information from the data, the key question that remains is
how —the other key question is
why, but I assume we all know our own
why, and I am quite confident we basically agree in the answer.
In my opinion, by painting a mask with a brush by hand, you are not taking your responsibility as an interpreter of the data. This is more a question of ethics and principles than a technical question. You can be extremely careful when you draw your masks, and you even can master it to the point that I can be unable to say whether you've applied a hand-painted mask or an algorithmic mask by looking at the final product. As I've said in my previous post, the final product isn't as important as the process for me. Or perhaps more than the process,
the reasons behind the process.
There's no love for the data when you paint a mask by hand. When you use the brush to sharpen a tendril of nebulosity by arbitrarily punching a hole in your mask, you're following an easy path that perhaps gives you the immediate answer that you (think that you) need, but you are not understanding what happens. The brush doesn't let you understand what are the relations between different structures in your data, and how the signal and the noise interact to yield the beast that an astronomical image is. In other words, the brush doesn't let you grow as astrophotographer. Ten years after this conversation, you can be using the brush exactly in the same way as you can use it today. The brush is easy, the brush doesn't make you questions, the brush will never fail or deceive you because it is the "ultimate" solution.
