Calibration travails - banding

[pvelez]

First post - So please let me know if I commit a faux pas here.

I have an issue with calibration. I seem to have banding in my images - I suspect that they are readout artifacts or else column defects that aren't being cleaned up in calibration. Or something else.

I've attached a screenshot of part of an image. This is 60 x 5 minute subs through a G filter. Taken with an STX16803. I have used the Batch Preprocessing script. I applied a master dark, master bias and master flat file. Optimise Dark Frame was checked. I didn't use the overscan option. The calibration frames were compiled using the process detailed here:

http://www.pixinsight.com/tutorials/master-frames/index.html

Note the horizontal bands.

What else can I report? Images were taken at -25C. I think I had about 40 subs for the master bias and 10 for the master dark. I can post those if that would help troubleshoot.

I tried to calibrate without bias - a pointless exercise as you'd expect as the flat wasn't properly applied.

Thinking the issue was with the master bias, I used the Superbias module to produce a Superbias frame. No joy there.

I'd appreciate any suggestions - I have some great data to play with but I fear my processing skills are letting me down.

Pete

[pvelez]

Doing some more reading and poking about the uncalibrated subs - they look like column defects. Because there are so many subs, they are being stacked which makes them more obvious the deeper I go.

Am I on the right path?

Pete

[Josh Lake]

If you go back and look at each level of the calibration, can you see which step causes the columns to appear? If you are not dithering your frames while shooting, this is likely a case of adding column defects, yes. You should try using Cosmetic Correction in conjunction with Batch Preprocessing, as shown in Harry's video here: http://www.harrysastroshed.com/pixinsight/pixinsight%20video%20files/pix%20videos/pixinsight%20cal%20mp4/pixinsight%20cal%20mp4.mp4

[sreilly]

That sounds right but I'm seeing them as more of a diagonal line and for column defects I would expect straight lines or am I seeing an optical illusion? It's just recently that I've noticed vertical banding in my images, a SBIG STL-11002M camera, and especially in the combined RGB images. Individually they are almost invisible but stacked and stretchered they become very apparent. I've seen some procedures for eliminating them but none that I recall for use in PI. I had bought a set of actions from Noel Carboni some years ago for PS and seeing how I already have that found they worked great. Not sure how they would help in developing a script in PI to do the same.

[bitli]

Banding is often caused by interference in the camera during the readout of the data. The interference may be due to bad filtering inside of the camera (either because it s used outside its specification or because of a defective component). They may also be caused by inadequate power supply, interference from other devices (computer, motor) via cabling or magnetic fields, grounding loop and other electrical problems.  The interference is in generally not synchronized with the camera therefore the oblique pattern. The interference may be very small but beat between the exposures after integration and so the result may have different frequencies and location than in the original images.  If you suspect that interference are the cause you may want to take a dark in normal imaging conditions (with motors running, at the telescope) and check if you can see any repeating pattern by stretching it to the maximum (they may not at the same frequency and position as the final one).  In that case you should strive to correct the source of the problem, which is easier to say than to do.

There may be other cause for banding I suppose (a nearby lighthouse  ?)

To remove them you could try to use the Fourier transform or maybe look how the CanonBanding script is doing it, although it works in a specific direction. Unfortunately I have no experience with these methods.

-- bili

[Carlos Milovic]

You may remove some of the high frequency banding with the  fourier transform / modification of amplitude component, but those strong interference lines won't disappear.  You'll have to get several frames (if dithered, best), and integrate them.

[pvelez]

If you go back and look at each level of the calibration, can you see which step causes the columns to appear? If you are not dithering your frames while shooting, this is likely a case of adding column defects, yes. You should try using Cosmetic Correction in conjunction with Batch Preprocessing, as shown in Harry's video here: http://www.harrysastroshed.com/pixinsight/pixinsight%20video%20files/pix%20videos/pixinsight%20cal%20mp4/pixinsight%20cal%20mp4.mp4

Josh, I think you have hit the nail on the proverbial head.

Looking closely, the bands are in the original frames. I've set up a defect map for each of the binning modes I use - x1 and x2 - and am now running the preprocessing script with cosmetic correction enabled. Will see how it goes and report back.

I assume I need a separate process for each binning mode.

Pete