StarTools

Ivo,
I have been struggling to improve my results and today I was able to bin my RAW Images 2 x 2 before calibrating and stacking. When I processed them in Startools I got better results. I believe that binning will improve SNR but I was also wondering if some of the algorithms in Startools are optimized for a certain image size? There are many settings in Startools that are pixel based and when I binned I had to decrease these (i.e. grain size in Wavelet De-noise) otherwise there was too much filtering. Can you give me your opinion what the idea image size is for Startools?...considering all of the default settings in your program. I could use this as an initial guess when making adjustments to pixel setting etc. for a much larger or smaller image. Maybe the ideal pixel scale for Startools would be more appropriate.

You may even want to consider having Startools store default settings per customer or maybe have the user enter a pixel scale so that these pixel setting could be expressed in arc-seconds.

Peter

Hi Peter,

Indeed, keeping everything else equal (camera, atmospheric conditions, exposure times, object, optical train), processing binned datasets will allow you to "push" the data more (see here about StarTools' software binning and considerations). This also changes the nature of the scale of the visible detail in terms of pixel sizes.

To be clear; StarTools is not by default more "attuned" to particular image or detail sizes - it simply assumes "perfect/flawless" signal + a Poissionian noise component.

More often than not, the only thing that necessitates a change of parameters is working around noise grain or simple personal preference. To make a long story short, when pixel size is mentioned for a parameter, more often than not is wholly data-set dependent (e.g. camera, atmospheric conditions, exposure times, object and optical train - dependent) or aesthetics-dependent. In neither cases can such settings be determined automatically or objectively - they vary per gear, per night and per user.

There are other cases where deviations from default values may be necessary/desirable. These almost always have to do with (consistent) imperfections in the optical train, stacking method/calibration or data acquisition techniques. Often (though not always), if you find an issue "goes away" at a smaller (binned) scale, then there are optimizations you can make with regards to the aforementioned.

I'd be happy to have a look at one of your stacks - there may be some quick pointers I can give you?