Strange artifacts in composite

Difficult to asses on the basis of the end result, but it looks like you have quite some light pollution or other artefacts in the subs? Can you show how the panels themselves look like?

Hmmm. I made no attempt to process the subs before trying to combine them. I'll remove light pollution on each of them and try again.

I don't think that will solve it as in the end the background is measured again and can be quite different. I seem to see artefacts in the panels in the background, just wondering if you have any issue in the subs themselves already, can you tell if you stretch a sub a bit much?

@vincent-mod No, no problem with the subs themselves--no artifacts like the lines just left of top center.

Are you saying the APP won't normalize the panels so differences in brightness are resolved? Other than alignment, that's the point, isn't it?

No, APP will do that, but if there are artifacts or just light pollution in the data, it will still be there. Maybe it's a good idea to share some data, because it's hard to see what it could be. Can you upload 5 lights, and 5 of each calibration file to our server?

Go to https://upload.astropixelprocessor.com and for the username and password, use the word: upload

Create a directory named “tailspin45-mosaicArtifacts” and upload in there. Thank you!

Thanks for taking a look! Five lights for four sessions uploaded plus master dark, dark flat, flat, and bad pixel map.

Thanks Tom, I'll have a look today!

Ok, so I see this is a mosaic (I think?), seems they're overlapping enough to make it a regular integration. Are you using any filters? If so, that will change my findings below;

Looking at your master-flat. It seems you could expose those a bit longer, the histogram for blue might be a bit low. I see you already expose at 10s, you may try to get an even light source that is brighter.

I tried an integration with all files at once, because you're using the same calibration files anyway, separate sessions are not needed. With this I got a warning from APP that you have some clipping of pixels going on, this is usually the case when you're exposing too short (you may want to expose for 3 min at least). That or there may be a small issue in the calibration data.

Looking at the result, I'm not seeing big issues. However, it does look like you may have used a Ha filter, if so, please indicate those details as that may change my findings.

Thanks for your effort and response.

Yes, this is a composite of 4 panes not a mosaic—there's plenty of overlap (maybe too much?). The uneven borders and artifacts are the problem.

Yes, I did use an L-eXtreme narrowband filter because I live in a Bortle Class 7-8 suburb of San Diego California. Which is why I use two-minute subs. I calculate the maximum actually should only be 1.4 minutes, although I just this moment realized that calculation assumes no filter. So yes, I'll try 3 minutes subs, at least. Thanks!

My flats (30) were exposed to about 28000 ADU. I can adjust the source colors to push up the blue and make the whole thing a bit dimmer so I get a longer expsoure with the same ADU values. Thanks for that, too.

I have about 250 two-minute subs for each pane, and APP crashes if I try to process more than one batch at a time—300 frames seems to be the maximum with 48Gb of RAM assigned in the config.

Curious what program you're using to get the pixel clipping info. Please don't say it PixInsight! 🤮 

Thanks again for the help. Looking forward to your advice knowing that I have about 1000 subs and used a filter.

Ah ok, so using a L-extreme. That will be mainly Ha and OIII. I'll reprocess by extracting that data first. I did see the borders still in the RGB processing I did, I think this is mainly due to the data itself. Let me get back to this today.

So extracting the Ha signal, you can tell that the signal is very low, this is the unstretched Ha histogram;

This is stretched:

Seems like the signal is not coming above background much, this will together with the darks and bias, clip pixels to black. In the end it will mean the integration is much noisier and this is likely what you're then starting to see in the borders as well even with proper integration. You'd want the peak of this signal to be shifted to the right so that you're not clipping the signal on the left-hand side. This would mean you need to increase exposure to 5-10 min per sub, possibly lower if you increase the gain a little. This is the downside of narrow-band filters, but you will get much in return at the same time because you're just collecting the Ha signal. Given your sensor isn't saturated at all, you can easily expose for longer.

So extracting just the Ha, does look better already. I only used your masterdark to get rid of the amp-glow. If you increase exposure, you'll have much better results for sure, without clipping issues. 🙂

Bit of light pollution correction;

Thank you very much, that all makes sense. Eager to try longer subs to avoid the black clipping.

Will report back for the benefit of other L-eXreme and Astro Pixel Processor users.

Tried 10 minute subs on the Seagull Nebula (IC1744), two nights, two panes each night.

The histogram moved right (barely enough to avoid black clipping), and the artifacts were no problem.

Very impressed with how APP handles composites.

Two more nights, two more panes will overlap the top.

Superb! Thanks for sharing!

Another night's work. Still need more data in the corners.

Local Normalization Correction helped at lot