Revised Feb 5th, 2022
Recently I noticed that, in my narrowband astrophotos, the red color of the stars in my calibrated LIGHT frames seems shifted a few pixels. The blue pixels are also shifted but by a much smaller amount. I decided to invent a visualization tool for CaLIGHTs that makes it easy to study this color shift.
In CaLIGHTs V3.1.6 I introduced a feature called SatGuard. This feature requires CaLIGHTs to be able to identify bright stars and determine what pixels need to be modified. I had already created a routine to identify stars and extract statistics which is used in what I call my CGEM Monitor. I decided to start with this routine and give it the ability to locate the precise location of these stars using a centroid calculation. It actually locates the precise location in three separate colors (Red, Green and Blue). These locations clearly showed that these precise locations were not the same for all three colors and the shift in location varies predictably throughout the image.
At the same time I decided to include a Half Flux Diameter (HFD) calculation. The HFD value is analogous to the Full Width Half Maximum (FWHM) calculation because both of them are useful for characterizing the size of stars in astrophotos.
I decided to make this new routine available in the next release of CaLIGHTs which will be Version 3.1.7. It’s called “Show Shift & HFD”.
The routine cuts out a small rectangle of pixels surrounding each star. The routine uses the green pixels from this rectangle to determine the “official” precise location of the star. The precise location is also determine using only the red pixels and finally only the blue pixels. As each location is determined, the HFD for each pixel color is also determined.
The small rectangle of pixels is then replaced with red, green and blue circular blobs. The size of these blobs indicates the HFD calculated for the red, green and blue pixels. The red and blue blobs are also offset away from the green blob to indicate their relative locations with respect to the green blob. I have exaggerated these findings so that they are easier to see.
If there is no significant color shift then all three of these colored blobs will be displayed on top of each other. The result is a white blob. If one color has a high HFD value then you should notice that it’s color blob is bigger. Here are some screen captures to explain what is going on here:
This is a close-up of the upper left corner of one of my calibrated LIGHT frames. It’s not easy to see but the red pixels are shifted slightly upwards and to the left. Now let me show you what happens when I click the Show Color Shift button.
1174 of the brightest stars were replaced with colored blobs. If you look closely these colored blobs indicate that the red color is significantly displaces towards the left. If I panned across the image this color shift would always get worse towards the edge of the image. Even though the red pixels are shifted to the left, the size of the blobs for all three colors are pretty much the same. This makes me thing that all three colors are coming into focus together.
The average HFD value is 5.86 pixels. This image is taken using my 8″ EdgeHD with a 0.7 focal reducer, OptoLong LeNhance Filter and my QHY294C camera. The pixel scale is 0.67 arc seconds per pixel so the Average HFD value is equivalent to 3.93 arc-seconds. This image was taken using a OptoLong LeNhance filter which has a narrow Ha passband. The following three images are what the pixel shift looks like in the corners of this image.
I also have a BK80ED Refractor. I examined photos where I used several different configurations of Focal Reducers and Filters. The images with the BK80ED do not show any significant pixel shift. Only the images taken with the 8″ EdgeHD + 0.7 Focal Reducer + OptoLong LeNhance Filter show the pixel shift.
What can Deep Sky Stacker do about this?
Once I realized this was happening, I got to wondering about what Deep Sky Stacker[DSS] is doing about this color shift. I know that DSS is taking all of my calibrated LIGHT frames and discovering where all the stars are located and then matching these stars up so that each LIGHT frame can be stacked.
Several years ago I started using a option in DSS which some beginners might have not notice or avoid using.
On the Results tab of the Stacking Parameters dialog there is a parameter called “Align RGB Channels in final image”. When this parameter is selected DSS does a very good job of characterizing the color shift of the stars. It then uses the characterization to shift the colors of the entire image.
As an experiment I decided to call up two versions of the stacked result for this astrophoto. First I stacked the images with this parameter not selected…
If you look carefully you can see that, even in the stacked result, the red pixels are shifted to the left and up slightly.
Clicking the Show Shift & HFD button makes the color shift obvious. Remember that I have exaggerated the pixel shift so that it is easier to see. This is a stack of 26 10 minute exposures. I can guarantee that when you try to digitally develop this image that the color shift will be very visible unless you apply some sort of transformation.
So…lets take a look at what this stacked result looks like when the “Align RGB Channels in final image” parameter is selected…
The red pixels look like they are not shifted as much. This image will be much easier to digitally develop because of the improved color alignment. Clicking the Show Shift & HFD button results in the following…
Wow! The alignment of the colors in the stars is much improved. Panning over the entire image shows that this is true for the entire image. The HFD value has increased from 6.79 to 6.84 which is a small amount. Obviously, the characterization that DSS uses is very powerful. This characterization also effects nebula and galaxies which improves the color detail in these objects as well. I don’t ever remember any issues that I could blame on using this parameter so I would say that everyone should enable this parameter.
For this post I had decided to display the average of the red, green and blue HFD values. I studied this a bit more and found that the three HFD values can be quite different. I decided to compare the FWHM values generated by Deep Sky Stacker (black pen) with the three HFD values generated by CaLIGHTs. I was pleased to see that the HFD values generated from the green pixels agree very well with the DSS FWHM values. I performed a Meridian Flip during this imaging run. Images 27, 28, 29 and 30 were 30s plate solving images. The first image taken after the flip was #31.
I think I will continue to look at these values. This imaging run was with the EdgeHD + 0.7 focal reducer + LeNhance Narrowband filter set-up which does exhibit red pixel shift. This graph says that the size of the star according to the red pixels is larger than for the green or blue pixels. I have a nagging feeling that this could also be a focus issue or even a backfocus issue…stay tuned.
Peter
Hi Peter ,
Just snooping around your Site and came across some new postings . Interesting addition to CaLIGHTs that you have implemented and planned .
I checked my Settings in DSS just to be sure and I did in fact have that Option enabled . I think I did this way back as I was coming from Planetary Imaging to DSO Imaging so I was aware of this option from using Registax . They have two Settings for this Feature…oops , seems like I can’t Post Pictures is this Comment Section . Oh well , they were called RGB Align and RGB Balance , so when I got to DSS and saw this Option I did select it .
Are you also aware of the “Debloom” Setting in DSS ? It would be interesting to see if the new Options in CaLIGHTs can get any useful info on whether this is actually having a positive effect or not . I use it and just convince myself that it works . LOL .
BTW , you mention earlier in this Article that you are using the “0.7 focal reducer/flatener” . If it is the 0.7X Reducer from Celestron then it is only a Reducer as the “Flattener” is built in to the draw tube in the EdgeHD by default , hence the ridiculous price for these things .
Looking forward to a new Season of Imaging !
Cheers ,
Scott Kuchma
Scott,
I use the RGB align because I did see some strange edge colours on my stars even with my Nikon D5300. I didn’t really understand how powerful the DSS align RGB function was until I created this post. I believe the RGB balance feature in DSS is actually a background calibration feature. If it is then there is a section in the CaLIGHTs help file that explains why I don’t use it and it explains the origins of the CaLIGHTs Sky Glow Comp feature.
As far as I know, the Debloom feature was meant for CCD camera which were problematic for blooming. Modern CMOS camera do not typically suffer from debloom. But like everything in astrophotography…unless you deliberately try developing an astrophoto with and without an option selected you will never know what helps or what doesn’t help. I even developed an astrophoto using every possible stacking mode and assess the noise in the stacked result to convince myself what works best for me. I might go back to those findings and post them here.
You are right that the EdgeHD 0.7 reducer is advertised as purely a reducer but Celestron does mention that it also preserves the flat field. I have a 0.85 reducer/corrector for my 80mm refractor and the corrector improved the flat field for imaging. All of these buzz words and descriptions are the least of my worries but it’s a fair point you make. Other amateur astrophotographers may be confused by my reducer/flattener descriptions so I will try to stick with what the manufacturer calls them…even though they both have flattening attributes.
Feb 28, 2022 I went thru this post and removed the flatener text for the EdgeHD focal reducer.