The order some these really good new tools are used in does matter.
As seems to be the norm these days, I was testing some workflow methods in PixInsight and some of the items in its giant toolbox.
I've learned some new things recently in the process of dealing with several issues in my work to reprocess IC63 project data from last summer. Those issues led to some very helpful suggestions from Russ Croman, the developer of many tools/methods used in PixInsight, the most prominent of those likely being NoiseXterminator, BlurXterminator, and StarXterminator.
What I've learned is that SPCC and SXT both need an almost untouched image for the best outcome. I learned that both tools use both the noise and the star PSF information within their algorithms. Processes that impact both or either of those image properties tend to result in less optimal outcomes when SPCC or SXT is applied. SPCC/SXT and similar tools use the noise/PSF data to tell the difference between stars and other objects so changes to those factors can result in stars being misidentified as the algorithms process the image.
Tools/processes that have that effect include drizzle (now a feature of WBPP), and noise reduction as examples of the most commonly used tools/processes that would have an effect. After many different options / tests runs in WBPP, I can verify that two things really have a big impact SXT when combined.
Drizzle has a noticeable negative impact on the effectiveness of SXT that becomes more pronounced as the star field density increases. IC63 has an extremely dense star field, resulting in a very poor outcome with SXT due to the drizzle I had used in WBPP. Russ Croman was extremely helpful when I contacted him, and his questions and suggestions led me to the issue caused by drizzle.
Reprocessing without drizzle and doing only DBE before processing the LRGBHa image in SPCC, then SXT gave the best result.
I tested that further in processing the Horsehead Nebula with it relatively sparse star field (as compared to IC63). I was able to verify effects from drizzle even with this object. I was able to test SXT's use at different points in the workflow; verifying that the later in the workflow it is used, the less robust result that was obtained.
In short, if your workflow is going to utilize SPCC and/or SXT they need to be implemented right after DBE for the best result.
One other 'change' to my workflow as a result of these tests is that I delay denoising/application of NXT until fairly late. For instance, for the Horsehead Nebula I did not apply NXT to the starless image until just before combining it with my RGB stars frame.
YMMV
As seems to be the norm these days, I was testing some workflow methods in PixInsight and some of the items in its giant toolbox.
I've learned some new things recently in the process of dealing with several issues in my work to reprocess IC63 project data from last summer. Those issues led to some very helpful suggestions from Russ Croman, the developer of many tools/methods used in PixInsight, the most prominent of those likely being NoiseXterminator, BlurXterminator, and StarXterminator.
What I've learned is that SPCC and SXT both need an almost untouched image for the best outcome. I learned that both tools use both the noise and the star PSF information within their algorithms. Processes that impact both or either of those image properties tend to result in less optimal outcomes when SPCC or SXT is applied. SPCC/SXT and similar tools use the noise/PSF data to tell the difference between stars and other objects so changes to those factors can result in stars being misidentified as the algorithms process the image.
Tools/processes that have that effect include drizzle (now a feature of WBPP), and noise reduction as examples of the most commonly used tools/processes that would have an effect. After many different options / tests runs in WBPP, I can verify that two things really have a big impact SXT when combined.
Drizzle has a noticeable negative impact on the effectiveness of SXT that becomes more pronounced as the star field density increases. IC63 has an extremely dense star field, resulting in a very poor outcome with SXT due to the drizzle I had used in WBPP. Russ Croman was extremely helpful when I contacted him, and his questions and suggestions led me to the issue caused by drizzle.
Reprocessing without drizzle and doing only DBE before processing the LRGBHa image in SPCC, then SXT gave the best result.
I tested that further in processing the Horsehead Nebula with it relatively sparse star field (as compared to IC63). I was able to verify effects from drizzle even with this object. I was able to test SXT's use at different points in the workflow; verifying that the later in the workflow it is used, the less robust result that was obtained.
In short, if your workflow is going to utilize SPCC and/or SXT they need to be implemented right after DBE for the best result.
One other 'change' to my workflow as a result of these tests is that I delay denoising/application of NXT until fairly late. For instance, for the Horsehead Nebula I did not apply NXT to the starless image until just before combining it with my RGB stars frame.
YMMV