PixInsight Forum

PixInsight => General => Topic started by: Starcruiser on 2018 March 25 06:19:06

Title: GREEN Channel Vertical Banding
Post by: Starcruiser on 2018 March 25 06:19:06
Since I using a dual-rig and camera setup for simultaneous acquisition I have run into a problem in processing only the Green channel.  Blue and Red combine fine even though the scaling is different PI has been handling this fine until now.  The vertical banding in the Green channel only appears while using the StarAlignment module.  Separate and apart the Green channels from the ST8300M subs and 168C subs are fine.  Once I align and scale the QHY168C green subs with the ST8300M green subs then QHY168C green subs become banded.

I have run out of combinations I can think of to localize this issue but have come at a dead end.  I do not understand why my process works fine for Red and Blue but Green gets messed up.  Until I can get this corrected I cannot use ImageIntegration on the green channel.

Attached is my PI desktop screen cap, Green channel is on the far left.  As you can see Red and Blue have no banding.

Please let me know what else I can post to assist with this issue.

Thanks,
Bill
Title: Re: GREEN Channel Vertical Banding
Post by: jkmorse on 2018 March 25 11:31:30
Bill,

First, that is more than vertical banding, but some kind of matrix grid affecting your Green channel.  Am I correct in assuming that you are star aligning each color separately?  If so, one thing to try is using the recommended method of star aligning all subs from all colors in the same run (its always best to minimize the number of times you run star alignment due to the distortions it can introduce).  Try doing that using a red or blue sub as the master and see what happens.

Best,

Jim 
Title: Re: GREEN Channel Vertical Banding
Post by: Starcruiser on 2018 March 25 11:58:35
Hi Jim,

Thanks for your response.

What I do is take all pre-processed RGB subs and use StarAlignment to register all the green subs for instance from both cameras with the reference image from the ST8300M for scaling the QHY168C to that image.  Same thing with the Red and Blue.  Everything worked fine with Red and Blue, just don't know why Green is misbehaving.  The relevant RGB masters are made from those stacks

I will try your suggestion and pick one of the Red subs as the reference image; I did not take luminance in this run.  And then run them all in one run.

Thanks,
Bill
Title: Re: GREEN Channel Vertical Banding
Post by: Starcruiser on 2018 March 25 12:37:09
Hi Jim,

I used StarAlign on all 66 subs using the ST8300M Red as the reference image to scale and align to in one operation.   Red and Blue subs are normal after registration and scaling but all the Green subs still have the same patterning. I have visually inspected all Green subs prior to the operation and there is nothing I can see in any of them to throw things off.

Earlier I did an experiment and registered all the QHY168C Greens together and they looked right.  Did the same to the ST8300M Greens and they looked right.  So I then took one ST8300M Red sub and used it as the reference image and then one QHY168C Green sub and just aligned those two images and the Green output had the same patterning as reported at the beginning.  Just doesn't make sense why Red and Blue outputs register and scale fine but Green is abnormal.  No warnings are posted in the Process Console.

In addition, another experiment:

Loaded one ST8300M Red and made it the reference Image, loaded one QHY168C Green for a target image in StarAlign.  If I use F5 and resultant Green image in the workspace is normal, if I use F6 global and resultant image in the output directory is patterned.

Starting to go nutz with this...ack!

Thanks,
Bill


Title: Re: GREEN Channel Vertical Banding
Post by: bulrichl on 2018 March 25 13:26:54
Hi Bill,

Please describe your workflow for the monochrome camera ST8300M and the OSC camera QHY168C separately in detail. Is it:

For the ST8300M:
- Calibration (each channel)

For the QHY168C:
- Calibration
- Debayer
- ChannelExtraction

Then StarAlignment?

Bernd
Title: Re: GREEN Channel Vertical Banding
Post by: Starcruiser on 2018 March 25 13:46:55
Hi Bernd,

Yes, it is exactly as you have stated.  Only exception is Cosmetic Correction for the ST8300M and then on to Star Align.

Thanks,
Bill
Title: Re: GREEN Channel Vertical Banding
Post by: jkmorse on 2018 March 25 14:58:43
Just for an experiment, try aligning the green subs WITHOUT calibrating them first.  I know that sounds crazy but I bet it "fixes" the problem.  If it does then we have a different direction to investigate.  I sometimes run into just this problem with narrowband images and simply have to do a work around of ImageCalibration by cleaning up the subs/image with CosmeticCorrection, ImageIntegration, and DBE.

Let me know if that helps at all.

Jim
Title: Re: GREEN Channel Vertical Banding
Post by: bulrichl on 2018 March 25 15:05:34
I am sorry, then I have no idea what goes wrong here. Hope that Jim's suggestion will be helpful.

Bernd
 
Title: Re: GREEN Channel Vertical Banding
Post by: Starcruiser on 2018 March 25 15:22:40
I am sorry, then I have no idea what goes wrong here. Hope that Jim's suggestion will be helpful.

Bernd


Yep this doesn't make any sense for me either.  Why can't PI process the Green just like it did Red and Blue instead of messing it up?

It worked great for this run:

http://www.kd0npt-astro.net/images/ngc281-lrgb_1435x1080_dualrig.html

Only difference now is I am using a QHY168C in place of the STF8300C on the 80mm scope.  Should only be a difference in the scaling calculations to me.


Thanks for the attempt!

Bill
Modify message
Title: Re: GREEN Channel Vertical Banding
Post by: Starcruiser on 2018 March 25 15:25:09
Just for an experiment, try aligning the green subs WITHOUT calibrating them first.  I know that sounds crazy but I bet it "fixes" the problem.  If it does then we have a different direction to investigate.  I sometimes run into just this problem with narrowband images and simply have to do a work around of ImageCalibration by cleaning up the subs/image with CosmeticCorrection, ImageIntegration, and DBE.

Let me know if that helps at all.

Jim

I can certainly  try that but I am so used to calibrating the OSC prior to debayering so I am not sure where that will lead us.

Title: Re: GREEN Channel Vertical Banding
Post by: pfile on 2018 March 25 16:31:34
the moire pattern / crosshatching effect in narrowband images can be due to the lack of background signal in narrowband images. the background is too near the dark signal and during calibration some pixels would be negative after calibration, and so are clamped at 0.

that can be avoided either by making longer exposures, or adding a pedestal to the calibrated files such that there are no negative values and thus no 0 values in the calibrated frames.

rob
Title: Re: GREEN Channel Vertical Banding
Post by: bulrichl on 2018 March 26 03:07:23
For "The Pacman Nebula" image no narrow band filters were used. Did the current problems occur with the use of narrow band filters?

Take a look at the histograms of the calibrated frames (monochrome camera) and the calibrated, debayered and channel-extracted frames (OSC camera). Compare red and blue vs. green channel. If the green channel of one camera is clipped severely, Rob will be right. In this case a wrong calibration (e.g. pre-calibration of the dark frames or the MasterDark) may be the cause. However, if the calibration was well, the measures that Rob suggested will put things right.

Bernd
Title: Re: GREEN Channel Vertical Banding
Post by: Starcruiser on 2018 March 26 04:47:02
the moire pattern / crosshatching effect in narrowband images can be due to the lack of background signal in narrowband images. the background is too near the dark signal and during calibration some pixels would be negative after calibration, and so are clamped at 0.

that can be avoided either by making longer exposures, or adding a pedestal to the calibrated files such that there are no negative values and thus no 0 values in the calibrated frames.

rob

No narrowband was used.  The OSC calibrated and debayered files were likewise combined with the RGB files of the mono camera albeit different image scales in StarAlign.  Exposures for both cameras was 300 secs.  Only the Green debayered channel acts up when aligned and scaled with the mono camera green channel.  Red and Blue do fine, so it is unexplainable at this point why Green cannot do as well with the same process.

Thank you for your input,

Bill
Title: Re: GREEN Channel Vertical Banding
Post by: Starcruiser on 2018 March 26 05:04:51
For "The Pacman Nebula" image no narrow band filters were used. Did the current problems occur with the use of narrow band filters?

Take a look at the histograms of the calibrated frames (monochrome camera) and the calibrated, debayered and channel-extracted frames (OSC camera). Compare red and blue vs. green channel. If the green channel of one camera is clipped severely, Rob will be right. In this case a wrong calibration (e.g. pre-calibration of the dark frames or the MasterDark) may be the cause. However, if the calibration was well, the measures that Rob suggested will put things right.

Bernd

Hi Bernd,

I am not using narrowband filters with this run either.  It looks at this point I may very well be better off starting out from the beginning and hopefully things will fall into place.  Don't want to tie folks up chasing one gremlin that may very well be operator error.  With the lousy conditions for image acquisition over the previous year I get infrequent use of PI, so it seems I am always starting over but I have had some successes with it.  Dual rig imaging does complicate things a bit but it does increase the yield for the available sessions.

Thanks for the input from  all.

Regards,
Bill
Title: Re: GREEN Channel Vertical Banding
Post by: bulrichl on 2018 March 26 05:11:10
The effect that Rob decribed is not confined to narrow band filters and can occur with RGB filters or OSC cameras as well. Did you check the calibrated frames as I suggested? If clipping is the cause, it will be obvious in the histograms (of the green channels).

Bernd
Title: Re: GREEN Channel Vertical Banding
Post by: Starcruiser on 2018 March 26 07:23:34
The effect that Rob decribed is not confined to narrow band filters and can occur with RGB filters or OSC cameras as well. Did you check the calibrated frames as I suggested? If clipping is the cause, it will be obvious in the histograms (of the green channels).

Bernd

Green did have an odd weak histogram so I ended up restarting at the beginning and all looks well.

Thanks,
Bill
Title: Re: GREEN Channel Vertical Banding
Post by: pfile on 2018 March 26 09:57:39
The effect that Rob decribed is not confined to narrow band filters and can occur with RGB filters or OSC cameras as well. Did you check the calibrated frames as I suggested? If clipping is the cause, it will be obvious in the histograms (of the green channels).

Bernd

yeah - i was actually commenting on jkmorse's reply, not necessarily trying to explain this situation. but i suppose whenever the exposure is short enough this problem can happen regardless of filter.

rob