PixInsight Forum

PixInsight => General => Topic started by: Mariner10 on 2019 February 16 19:58:38

Title: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 16 19:58:38
Hi all,
Attached are images of the Horsed Head, one is after cosmetic correction has been applied and everything looks okay the other is after the debayering process has been applied. The problem should be obvious.
I've never had this occur before and I was hoping subsequent processing would remove the lines and gradients (if that's the correct terminology).
All images were taken with a OSC QHY8L, calibration files added.
The debayer matrix for that camera is GBRG which was applied and has successfully been applied in the past.
I tried all other debayer combinations but they all gave either the same result or made the same result worse.
Blink did not give any indication that there was going to be an issue and even when I applied the debayer process to just one sub the same result occurred.
Is there a problem here or will this effect be removed later in the processing procedures. I have already tried DBE and applied ABE afterwards both of which took out the vignetting but the lines are now more obvious.
Regards, Steve.
Title: Re: Introduced lines into debayered images.
Post by: pfile on 2019 February 16 22:41:23
i think you need to stretch those and re-post them because they are so dark i can't see any of the artifacts you're talking about...

rob
Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 17 15:07:18
Sorry Rob, I forgot that they don't save stretched. I'll repost it when I get a moment.
Cheers.

Steve.
Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 19 00:23:26
Hi Rob,
So I've uploaded the stretched images (I hope) over two posts as I keep getting an error message that they are too large. You should be able now to see what I'm on about.
Cheers.
Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 19 00:24:09
This is the second image Rob.
Title: Re: Introduced lines into debayered images.
Post by: pfile on 2019 February 19 09:32:00
going to guess that these were registered and stacked and then debayered?

debayering needs to happen to each sub prior to registration.

rob
Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 19 14:30:20
Hi Rob, no, the process was cosmetic correction of all subs, debayered all subs and then registered them hoping that ABE or DBE might help resolve the problem.
I might have to go back to square one and start the process again. Even if the debayer matrix was wrong I don't think it should have done that to the images.
Cheers.
Title: Re: Introduced lines into debayered images.
Post by: pfile on 2019 February 19 15:47:28
well, this shouldn't happen obviously. maybe the CC step has somehow messed up the images. worth omitting that to see what happens.

rob
Title: Re: Introduced lines into debayered images.
Post by: chris.bailey on 2019 February 20 00:28:43
Would help if you could dropbox one of the baseline .fits files.

Chris
Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 20 18:36:32
Hi Rob,
I'll try giving the CC process a miss and see what happens then.

Hi Chris,
I'll drop one of the original subs into dropbox over the weekend. It's going to be stinking hot here so a perfect time to sit in the AC and do some processing.

Cheers.
Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 22 16:16:42
Hi Chris,
The link to the dropbox is https://www.dropbox.com/s/0tv07xkz7oiaa1f/HorseHead-1-600s-8-C0.fit?dl=0
Hope that works, I haven't used it before, if not let me know.

I'll try your suggestion this afternoon Rob.

Cheers.
Title: Re: Introduced lines into debayered images.
Post by: pfile on 2019 February 22 16:39:20
this isn't straight out of the camera though, is it? the FITS header indicates it was written out by Pixinsight.

the striped pattern appears to be in this image even before debayering... something is very weird here. plus it looks like it was saved with overscan areas preserved?

re: earlier w/ no problems vs. now, is it a new version of the capture software and/or the driver for this camera that differs between then and now?

rob

Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 22 18:34:35
Hi Rob,
No, the camera saves it as a fits file, I just checked all my other images that are waiting to be processed and they are all fit files if that's what you mean.
These were taken in 2016 and I haven't, to the best of my knowledge, updated the software for the camera. The laptop I use is kept with my gear and I only image remotely where there is no internet available so no auto-update availability.
I might try and see what happens if I try to process subs from last year and see if I get the same result.
Cheers.
Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 22 18:36:34
Forgot to add, I tried your suggestion of debayering before cosmetic correction was applied with the same result so it appears that the fault is right at the beginning. I took a single original image and applied debayer to it with the same result.
Title: Re: Introduced lines into debayered images.
Post by: pfile on 2019 February 22 22:06:32
well what i'm saying is that the file you posted seems to have been read into PI and then saved from PI, since the FITS header says:

 PROGRAM = 'PixInsight 01.08.06.1457' / Software that created this HDU           
 COMMENT PixInsight Class Library: PCL 02.01.11.0938                           
 COMMENT FITS module version 01.01.05.0433                                       

your capture software probably would not put those FITS keywords into the header.

something is funny with these images, i just don't know what. it's unusual to have those black borders all the way around straight out of the camera, and the lines seem to be there even before debayering...

rob
Title: Re: Introduced lines into debayered images.
Post by: bulrichl on 2019 February 23 01:44:48
The sensor of the QHY8L has dimensions of 3110 x 2030 pixels. Your image has a width of 3040 and a height of 2016 pixels. Nevertheless your image has dark borders (however they are not black!), the effective image has only 2777 x 2010 pixels. This is very odd.

As Rob noted, the FITS file was written by PixInsight and not by a capturing software. I can imagine two different possibilities:

1) An unsuitable driver was used for your camera or
2) the image was already processed in PixInsight.

One needs a subframe that is stored by the capturing software in order to analyze the issue further.

Bernd
Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 23 17:07:40
Well the mystery just deepens because these images have never been in PI, they just sit in a folder on my laptop waiting to be processed.
I'm thinking that I might just need to bin the lot and wait until next summer although March in a couple of weeks is an option.
It's frustrating that's for sure.
So if you're out of options then I think I might just do that.
Cheers.
Title: Re: Introduced lines into debayered images.
Post by: bulrichl on 2019 February 24 01:26:28
I am talking of the original images out of the camera - we would need to see one of those. These files were saved by some software. Perhaps you could tell us which capturing software did you use for capturing (e. g. APT, SGP, or software of the camera manufacturer...)?

Bernd
Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 24 18:54:07
Hi Bernd,
The image I dropped into dropbox was an original, straight from the camera which is a QHY8L using EZCAP as the preferred software for that camera.
The images were taken in 2016 (from memory as I don't have them in front of me at the moment) but now that I think about it I had an issue with calibration files not being the same geometry as the original images so that is more than likely why the the FITS were written by PI.
So I would have tried to calibrate them without success so was forced to alter the dimensions of the originals to that of the calibration files. I wouldn't have thought that would cause the lines but it might explain the dark edges.
Sorry, I should have remembered that but the issue still exists at the very start of the processing procedure.
Cheers.
Title: Re: Introduced lines into debayered images.
Post by: pfile on 2019 February 24 21:42:30
the image you posted *can't* be straight from the camera because the fits header indicates it was written by pixinsight...

anyway if indeed you had to alter some of the calibration images you may have cropped them on a non-bayer matrix boundary and that can definitely screw everything up.

rob
Title: Re: Introduced lines into debayered images.
Post by: chris.bailey on 2019 February 25 00:36:08
If you downsample that image by a factor of 2 (to destroy the bayer pattern) the vertical pattern is still there so I think it happened at capture time. Electrical interference, low voltage to the camera, wrong driver.

Title: Re: Introduced lines into debayered images.
Post by: Mariner10 on 2019 February 25 16:06:23
Thanks guys, if that's the case then I'm forced to bin the lot and try and get the object again in a couple of weeks time.
Appreciate again your time and efforts.
Cheers.