PixInsight diagonal lines weirdness

[BrendanC]

Hi all,

I'm completely at a loss as to what's going wrong here.

I've got a load of RGB data, and Ha data, from an OSC camera (ASI533MC Pro). 

Within PixInsight, I stacked all the RGB, and all the Ha, using WPBB. Then, I've extracted the G and B from the RGB, and the R from the Ha, aligned them using Star Align, and used Channel Combine to bring them together with the Ha for R. All good so far.

HOWEVER, when I start stretching, I get this bizarre set of parallel, diagonal stripes. I cannot figure this out at all. None of the individual channels show this. The example image is just combined, stretched in GHS, and then some curves applied to show what I mean. Any ideas? I'm stumped. I've added another version with yellow lines to emphasize where the lines are.

 

[scitmon]

Might help to post your stacked individual channels here for us to look at.

I find it hard to believe that channel combination would introduce these artefacts.  Try LinearFit before channel combination?

Edited January 4, 2023 by scitmon

[BrendanC]

I'm pretty new to PI so it's entirely possible I'm doing something wrong, although I'm familiar with DSS, Siril and APP so I know about stacking. In fact, I'm stacking the files right now in APP to see whether there's any difference.

Would be great if you'd like to take a look: https://1drv.ms/u/s!AqovBuVZMwj3mJIg9H5vuCW4cXpa1Q?e=Fs8Imm

That folder contains the XISF and separate TIF files, so just grab what you want.

I'd be very interested to see if you can replicate this, and if you can get it to work, how you did it. 

Thanks!

[scitmon]

OK I had a quick go with your stacked files and here is my result:

I don't know if those artifacts are hidden in my image or don't exist.  I can't explain how you got them.  

My workflow was:

- ran AutomaticBackgroundExtractor on both ha and rgb (subtraction)

- extracted red from ha, and g and b from rgb

- staraligned ha_r to rgb_g

- ran LinearFit with rgb_g as reference on ha_r and rgb_b

- ChannelCombined them

- did a STF stretch via HistogramTransferFunction

 

Perhaps someone else can shed light on the mystery.

 

Image14.xisf

Edited January 4, 2023 by scitmon

[BrendanC]

Thanks for doing this, really appreciate it.

I just stretched your XISF file and cannot see the diagonals! This is interesting. The only step I haven't done is LinearFit, so I'll look into that.

In the meantime, I've stacked the files in APP and I don't get the artefacts. So, that implies acquisition isn't a problem.

But, when doing that, I think I might have found a stray flat file hidden among the flats that might be causing an issue. So, I've got PI back on the case, currently restacking, will try Linear Fit, and see how I get along.

Thanks again!

Edited January 4, 2023 by BrendanC

[edarter]

How are you finding WBPP Brendan? I've not used it yet, can't drag myself away from APP for stacking as its what i've got used to, and to be fair is very very good!

[BrendanC]

I've only ever stacked in PI using WPBB, but I like it.

I watched the full, free Adam Block WPBB intro series on YouTube, and he says somewhere in it that there is no difference whatsoever in the result, whether you do it manually (the hard way) or using the script (the easy way).

It's initially quite intimidating, like everything in PI. There are some foibles I still don't like, for example always having to tell it to debayer lights, and it takes a very, very, very, very, very long time to stack (even longer than APP believe it or not), but other features are great, mainly the ability just to load files, and it reads the FITS headers, and assigns accordingly. So, no need to upload bias, darks, dark flats, lights etc separately, just chuck it all in and provided you've got the right types in the header, it all automagically sorts everything out. You can even point it to a folder and it will pick up everything in there and do its thing.

This means you could, if you wanted to, load up your entire bias, darks and dark flats library, and it will automagically assign the right ones to the flats and lights for that session. In fact, that's what I've done, except I only use bias frames now after lots of comparing (thank you ZWO for the amp glow reduction circuitry in the ASI533MC), so I just keep the master bias in place, load up lights and flats, and go. Very handy.

 

Edited January 6, 2023 by BrendanC

[edarter]

Interesting, I will have to have a look at this.saw a video the other day by Sascha Weiss that said to use APP over PI as its stacking is better. I did think at the time he would be ripped to pieces over that lol.

[teoria_del_big_bang]

12 minutes ago, edarter said:

Interesting, I will have to have a look at this.saw a video the other day by Sascha Weiss that said to use APP over PI as its stacking is better. I did think at the time he would be ripped to pieces over that lol.

I think at one time this was true, I had seen the same recommendation some years ago but a lot of work has gone into this area in PI and would think there is now no reason not to use PI for all pre-processing (someone will correct me if I am wrong here).

Personally I really suggest you use WBPP script and at least have a compare. It is so easy to use especially when you grasp the use of keywords and for any image using multiple sessions with different flats or different exposure times it is a Godsend.

Steve