NormalizeScaleGradient Pixinsight

[AstronomyUkraine]

I finally had chance to test out the new NormalizeScaleGradient script in Pixinsight, with some broadband data. I processed the red channel from data collected on IC 59 and IC 63, with my normal workflow, which is SubframeSelector - Align - ImageIntegration. Followed that up by using the NSG workflow, which is Align - NSG - ImageIntegration. 

 

This was a basically a head to head between SubframeSelector and NSG, and the results were inconclusive. The reference frame I used in NSG was the same reference frame SubframeSelector suggested, which as it turns out was also one of the frames with the highest alt reading, the main criteria for choosing a reference frame in NSG. Both SubFrameSelector and the NSG script discarded no frames at all, which suggests my stars were nice and round and had no problems in them.

 

As you can see from below, I have two sets of data, the red image, the red ABE image, the red background image, followed by the 3 equivalent NSG images. I can't see much difference between either sets of images, bearing in mind my light pollution here is Bortle 4, so I don't expect much gradient. People living in more light polluted areas may find the NSG script performs better than using the SubframeSelector.

 

Red stack SubframeSelector

 

 

Red NSG

 

 

Red ABE

 

 

Red NSG ABE

 

 

Red background

 

 

Red NSG Background

 

[TerryMcK]

Great review Brian

[AstronomyUkraine]

17 minutes ago, TerryMcK said:

Great review Brian

Thanks Terry. Like I said the result was inconclusive, but at the end of the day, the final image is the one that counts. A better test for this NSG would be someone shooting broadband in London, or some other heavily light polluted location.

[MarkAR]

I think your experiment has just shown that good data is good data no matter how it's weighted. 

 

I lost 17x 600s of Oiii on my last image through NSG, combination of focus error, moon and hazy clouds.

[AstronomyUkraine]

6 minutes ago, MarkAR said:

I think your experiment has just shown that good data is good data no matter how it's weighted. 

 

I lost 17x 600s of Oiii on my last image through NSG, combination of focus error, moon and hazy clouds.

If you run that same data through SubFrameSelector I doubt it would discard so many, unless they were really bad. The only ones I usually get rejected are for eccentricity.

[MarkAR]

They were really bad 😆