Star Intensity During Stacking

[shurik]

Hi Community,

This issue might have been discussed before, but I am dealing with a very common problem how to control star shape, intensity etc. during image stretching and processing. One useful approach is to work on the "starless" image first and then bring in the "star field" when everything else is done. while this method has its own limitations and introduces a number of artifacts (which could be the topic of separate discussion) my thought was why not to start generating "starless image earlier during stacking steps. Could there be a plugin or an algorithm implemented into stacking process that detects every star in the image and does not add the signal in that region from the next image or even better makes it less intense (some sort of rejection above certain threshold). This way final stacked image will have less  prominent stars that will be easier to deal with or generate "starless" version that is suitable for further processing steps.

Any thoughts, tutorials or references if this has been implemented in some form or shape would be very desirable.

thanks, Alex

[eganz]

Alex,

have you tried masked stretch?
This should help maintain the color, shape, etc. of the stars

during the stacking, we are not losing any information by adding/averaging together all of the stars… It's all in float. there is however often a lot of noise at this point, making it a bad time to remove/separate stars before stacking is finished.

Eric

[shurik]

well my idea in terms of star signal was as follows: first we are adding it (during stacking which often results in very bright stars that overpower the rest of neboulosity) and then we are removing it (during creation of starless images or masked stretches, or star layer removal  etc etc) so why not to deal with it from the beginning: simply not add it during stacking or even reduce it during stacking?

[Juan Conejero]

Hi Alex,

Sorry for being so sharp, but several design principles of PixInsight go in the exact opposite direction to this. Image integration has nothing to do with custom post-processing techniques, such as star masking or 'star removal' (the latter being very questionable in my opinion, BTW). Mixing these concepts leads to the creation of automagical black boxes. While there are other softwares out there that center their development around such devices, we definitely prefer rigorous implementations in PixInsight.

[shurik]

Juan, I understand your point, and it was just an idea that i thought cold be implemented. I believe 99.9% of posted images have manipulated stars in one way or the other (mostly size reduction, intensity reduction, color enhancement etc) so there is no "scientific value" in any case once the image becomes non-linear, but thank you for clarifying, I respect your design principles for PixInsight