ML7087373 wrote:The linked image is a 92 MB image of M33. It is a DSS stack of 11x180 second lights, and 3x180 second darks. The image is captured on a Canon 60Da at prime focus on a TEC 140 APO ED (f/7). ISO is 800. There was a good deal of light pollution and moon light. The images are unguided, but there is no trailing as my mount was well polar aligned.
This is the second image I've processed with ST, and now I have enough experience to ask questions. If you would be so kind please do your best with this image and post your tracking history so I can repeat it to see the effect of the changes.
https://docs.google.com/file/d/0BzTNnHs ... sp=sharing
Thanks!
Before I begin, could you tell me whether you shot in RAW mode? The image seems somewhat stretched and possibly noise reduced by the in-camera JPEG processing engine. I also find it hard to bring out any faint detail (possibly because that detail has been destroyed by the JPEG engine and the low 8-bit JPEG dynamic range). Processing a stcak from a JPEG source is very hard and will lack a lot of detail!
The first thing I notice is that post autodevelop I end up with an odd orange image. Why is that?
AutoDevelop does its absolute best to bring out any things that it thinks are the most important in your image, by allocating the most dynamic range to things that it thinks stand out. That's why it is the absolute first thing that I do - it's a perfect way to assess any issues in your image - it's very good at showing you things like; a bias (for example caused by light pollution), stacking artifacts, noise grain, gradients, dust bunnies, etc. Once those issues are dealt with, it will, in the absence of any of the previously mentioned artifacts, start working on the 'real' cleaned up data and find the best histogram transformation curve (also known as 'levels and curves' in Photoshop speak) for your image (or part of your image if you specify a particular 'Region of Interest').
In StarTools, you can redo your global stretch as many times as you like, even if you have performed operations after. StarTools will 'go back in time', apply your new global stretch and re-apply everything that you did after you did your previous global stretch. This is how you can use AutoDev (or it's manual cousing 'Develop') multiple times.
This 'time traveling' feature is an integral part of StarTools and is the reason why it yields better results, yet is much easier to operate - the sequence in which you do things is much less important in StarTools as operations travel back in time and insert themselves into the timeline as necessary. For example there is no distinction between linear on non-linear data and therefore no distinction between operations that can only be performed on linear or non-linear data. Good examples of the latter are deconvolution or color calibration - in StarTools you can apply deconvolution any time you like, even on non-linear data, while color calibration (a linear thing) is preferably done at the very end on the end-result, rather than at the beginning.