First LRGB image - strange RGB glows around edges

[Elp]

Im not saying it's impossible to take flats afterward as many people do, but for someone who's fairly new to the process it's good practice to lay a good workflow from the start. That is why I recommend to take them at the time. If we shoot in a climate which has fairly stable outdoor temperatures with minimal temperature fluctuations flats can be taken before or after lights, with a cooled camera it's even easier to do. As the individual gets more experienced they can change the process to suit their requirements. The OP has a strange issue and we're trying the recommend the best course of practice to eliminate any possible errors which could be contributing to their issue. I have to assemble and dismantle my setup every session so for me doing what I've recommended works best for me so I've passed that experience onto the OP.

[BrendanC]

OK, it's happened again. I'm nearly at my wits' end here.

This is what I get for the Bode's and Cigar galaxies - stacked in APP...

... and stacked in DSS.

(Yes, I know there are sat trails in the DSS version, and possibly walking noise in both, these don't bother me because I know the fix - dither more frequently, use a different algo in DSS)

Both processed in StarTools, using pretty much the default settings, no noise reduction applied. I haven't tried getting 'the best' result, because I can clearly see these images are not right. I can process pretty much any other set of data I took with my DSLR using the same default settings and they're OK.

This is a high, easy target which I've shot before with my old DSLR, no problem. No bright lights nearby, Bortle 4 sky, no wind, no clouds, no Moon.

I've checked that the right filters are in the right slots, the right way around (they're mounted, so it's impossible to get them the wrong way round because they screw in), and that the right filters are selected in APT, by literally watching the correct filter appear when I select it manually.

I'm going to try and provide as much detail as I can here, in case anyone can help. I've tried to (re-do) everything by the numbers here. I really do not know what I'm doing wrong.

• All lights and calibration frames are shot unbinned, at gain 139, offset 50, temp -15C (some of the Luminance flats were between -11C and -15C)
• Taken over two nights, but with everything absolutely untouched between each shoot and virtually identical sky conditions
• Luminance is 60s per sub, RGB is 180s
• Total integration time is L 5880s, R 2520s, G 1980s, B 2340s
• Flats were taken on the first night, in astro darkness, using an LED screen, with a perspex sheet and some sheets of paper, after using the APT flats tool to calculate exposure times
• Dark flats were taken last night, again in astro darkness, using the exact same plan as for flats so that all details matched up, with the scope on the mount, cover on, a hat over that, another hat over the primary mirror end of the OTA, and two thick towels over everything
• Darks were then taken in the exact same conditions last night, for 60s and 180s exposures
• 25 flats per channel, 25 dark flats, 50 darks per sub length
• Kit list is 130PDS, NEQ6, ASI1600MM-Cool, ZWO EFW with 1.25 inch mounted ZWO LRGB and Baader HSO filters, Sky-Watcher 0.9x coma corrector, Sky-Watcher auto focuser with HitecAstro DC Focus V2, APT, PHD2, APP and DSS to compare stacks, StarTools

Questions are:

• When I was checking my filters, I noticed that my ZWO L filter is actually called UV/IR cut. I think this is the same, but could someone confirm?
• I have the 8-wheel EFW and just 7 filters, so position 8 is empty. Could this be causing any problems?
• Am I definitely right about the filter wheel being such that the EFW lettering is on the camera side, and the screws on the coma corrector side?
• I've read that the older ZWO LRGB 1.25 inch filters may suffer vignetting, especially with faster scopes. With my 0.9x coma corrector, I'm at around F4.5. Could this be an issue?
• I did get a calibration warning in APP while stacking, but when I retried the exact same files - as in, literally just pressed the integrate button again with the same file list in place - it worked. I've posted about that here: https://www.astropixelprocessor.com/community/main-forum/critical-warning-flat-field-calibration-2/ Could this be a pointer to what's going wrong? DSS didn't spot any issues.
• I know that the LED screen approach to flats isn't the best, but I know other people have said they use it successfully. However, again, could the flats be at issue? They worked fine for narrowband, and with my DSLR. How would I check? Should I try something like sky flats or the t-shirt method and see if that yields better results?
• When I'm processing in StarTools it seems fairly obvious to me quite quickly that there's heavy vignetting going on. Could that be the flats? Or, could it be the filters?
• Given that some of my Luminance flats were between -11C and -15C, surely that wouldn't have such a dramatic effect, would it?
• Finally... deep breath... given that I'm losing what little hair I have left here, would any kind soul (@vlaiv?  @Elp? @Varavall?) be willing to take a look at the entire set of data? If so, it's here - subs, flats, dark flats and darks, total 13.4GB unzipped: https://1drv.ms/u/s!AqovBuVZMwj3ku8zByEPVWYQIHSeLQ?e=NaPSm2
  
  ... or 5.7GB zipped in separate folders (I couldn't get the whole thing to upload in one big zip file): https://1drv.ms/u/s!AqovBuVZMwj3kvMCfUjIHVMBAxfbvQ?e=5QLric

Actually, finally finally, what do I do if the camera is borked? It's second hand so I don't have much recourse if so, and I certainly wouldn't be able to sell it on. And if I did get, say, an OSC instead, and there's something wrong with my technique, then I'm back to square one.

My workflow was totally sorted with the DSLR, so I thought it would be a fairly smooth transition to LRGB. I got the hang of narrowband with this camera pretty quickly, so perhaps it's not the camera, it's me:

If you've managed to read this far, then thanks!

TL:DR
I'm going insane trying to get my new camera to work and if you could help, I'd really appreciate it.

Cheers, Brendan

Edited April 3, 2022 by BrendanC

[Varavall]

32 minutes ago, BrendanC said:

Finally... deep breath... given that I'm losing what little hair I have left here, would any kind soul (@vlaiv?  @Elp? @Varavall?) be willing to take a look at the entire set of data? If so, it's here - subs, flats, dark flats and darks, total 13.4GB unzipped: https://1drv.ms/u/s!AqovBuVZMwj3ku8zByEPVWYQIHSeLQ?e=NaPSm2

I don't have much planned for Sunday, so I'll take look tomorrow and see what I get. Should be an interesting exercise!

[BrendanC]

Superstar. Thank you. 

[Varavall]

Of course it may be next Sunday by the time the download finishes!!! I live in the mountains so rely on a 4G router; no fibre here.

[tomato]

Hi Brendan, I use APP for my calibration and stacking and StarTools quite a bit so I will give your data a go.

[BrendanC]

2 hours ago, tomato said:

Hi Brendan, I use APP for my calibration and stacking and StarTools quite a bit so I will give your data a go.

Hey, thank you!

2 hours ago, Varavall said:

Of course it may be next Sunday by the time the download finishes!!! I live in the mountains so rely on a 4G router; no fibre here.

Yikes. There is the zipped version too, but when I tested the unzipped download it seemed to have zipped everything up anyway so that may not make a difference.

[Elp]

@BrendanC I downloaded your data, and was dreading finding something amiss. Doing a quick stack of each L, R, G, B set separately in DSS, and then opening in PS to do minor level adjustments and a few contrast tweaks (ive made them a bit too bright but was doing it quickly), your data seems like what I would expect from an LRGB image:

The colour gradients are there still but not blown out like the images you've uploaded which would suggest a post processing issue. I'll wait and see what other find with the data.

I did notice with the lights as you've taken them on more than one occasion some have done a meridian flip but your flats were all taken at the same time in one session hence one orientation.

Edited April 3, 2022 by Elp

[Varavall]

1 hour ago, BrendanC said:

Yikes. There is the zipped version too, but when I tested the unzipped download it seemed to have zipped everything up anyway so that may not make a difference.

Finally it all arrived over 4 hours. I have stacked the images and calibration files using ASTAP and aligned and processed the stacked subs in GIMP. There is a good amount of fine detail around the galaxy, but I think you need to take more subs to build that fine detail. Of course you could stretch it until it blows out the galaxy centre. I couldn't recreate the kaleidoscopic effect without extreme stretching, i.e. overprocessing the data. Below is what I achieved in about 15 minutes in GIMP (exported jpg), but I think I have demonstrated what I wanted without spending too much time on it. Maybe the rich people here with PI and PS can pull out more detail! Interesting exercise and I hope it helps.

Adrian

[symmetal]

Brendan,

I'll process your files later with APP and Startools to see how I get on.  You've applied far too much saturation in your postings though that may have been to indicate the colourd patches.

Alan

[vlaiv]

@BrendanC

Have you tried what I suggested to see if there is light leak in your setup?

I just downloaded one of each subs and did simple calibration and pattern is present in calibrated sub - but it does not correspond to any of calibration files - which suggests that it is not calibration related.

So here are in order - very stretched dark, that looks like usual dark from ASI1600, calibrated light frame, and single flat frame.

To my eyes pattern in lumianance sub after calibration does not "carry" signature of either dark of flat.

It sort of has darker interior with dominant dark patch being around M82.

Flat does not have such feature. It does - but it is located in other place. Peak intensity is closer to M81, and if flat is under or over correcting - we would be bright or dark spot near M81.

Similarly - if we compare patterns in dark - they don't match background in calibrated sub.

If you look at result of your stacking - you'll notice something interesting

This is image split by color channels.

It looks like light leak is changing position as scope tracks across the sky. In fact - Red and Blue (first and last) - look like 180 rotated images which suggests that you did red and blue channel on different sides of meridian and that scope was flipped between these two filters with respect to light source that cause the leak.

 

[BrendanC]

2 hours ago, Elp said:

your data seems like what I would expect from an LRGB image:

Thank you so much for doing this. I really, really appreciate it.

1 hour ago, Varavall said:

Interesting exercise and I hope it helps

Definitely, and again, thank you for taking a look.

40 minutes ago, symmetal said:

You've applied far too much saturation in your postings though that may have been to indicate the colourd patches.

I've just run the image through StarTools with default values, which usually works. Having said which, I've managed to get something... see below.

@vlaiv I must admit, I'm struggling to get my head around the various analyses you've done, but to summarise, you're saying that there's an issue with the data capture, not calibration or processing? And that this issue is a light leak of some sort? I have been taking some shots in broad daylight, with the ends of the scope covered but the camera/EFW/adaptors/focus tube uncovered, and I cannot see any obvious light leakage. The primary end, while I'm shooting, always has a waterproof hat over it to avoid this. Regarding your other advice, given that I have what I would call 'proper' darks now - taken with the scope completely covered, in astro darkness, without the remote possibility of light getting in anywhere - is the idea now that I take a dark in astro darkness but without covering the whole scope, just with the cap on? Would that indicate leakage? I'm not quite sure exactly what I should do.

Finally, I've been getting help on the StarTools forum too, and this is where I'm at - but with quite aggressive gradient wiping which I don't feel I should be doing:

The slight problem now is that I took a quick set of subs the day I got the camera, completely uncalibrated, and just quickly stacked them and removed the vignetting in Affinity, and I got this which, to my eye, is better! So, I'm getting somewhere, I'm just not quite sure exactly where yet...

Thanks again everyone, I'm really appreciative of your help.

Thanks, Brendan

Edited April 3, 2022 by BrendanC

[BrendanC]

Latest version from an APP stack (which takes about 100x longer than DSS). This is another improvement, but again, I'm wiping the gradient very aggressively. @vlaiv, I'm definitely going to give your diagnostic recommendation a go, but if you could just clarify for me what the next step is, to compare my darks with, that would be great.

[vlaiv]

30 minutes ago, BrendanC said:

I'm not quite sure exactly what I should do.

Here is what I would do to diagnose light leak:

- take your setup in conditions you would otherwise use to shoot lights and just cover the scope with front end cap and shoot single or small sequence (in fact you should take 3-4) of darks at exposure lengths you are usually using (like one to few minutes). You can do this on a cloudy night so you don't waste any imaging time.

- Take your camera off the scope and bring it into a dark room. Put cap on it and place it face down on a wooden table and take same darks as you did on a scope (so same exposure, gain, offset, temperature - all).

Post subs for inspection, or do inspection yourself:

- different mean ADU value between subs is red flag

- any sort of pattern in sub created by subtracting two darks is red flag.

Ideally you want their mean ADU value to be the same and when you subtract one from another - result should have ADU of 0 (will be the case if mean ADUs are same) and stretched result needs to be pure noise - no pattern present what so ever.

If you find above not to be the case - cause might be light leak. If you do the same with two "scope" subs and two "house" subs - and you find that you don't get the same pattern - it is 99.99% light leak.

Dark subs need to be 100% the same except for the random noise - if they are not, it means that there is some sort of external signal in them that is changing with external conditions (like taking subs on scope or in house).

 

[BrendanC]

20 minutes ago, vlaiv said:

any sort of pattern in sub created by subtracting two darks is red flag

I'm not sure what you mean by subtracting one from the other. If I were to open them in, say, Photoshop layers, does this mean I would use the Difference or Exclusion blending mode for example?

Also, given that I'm sure the darks I took in the garden are good - I mean, the scope was swathed in towels, including the camera and focus tube, the cap was on, there was a hat over the cap, and over the primary end, under the towels, all in astro darkness - then I should be fine classing them as the 'house' versions, right? And then I can do the thing you suggest, of taking equivalent darks but with just the cap on, and use them to run this analysis? Or to be on the safe side, should I retake some 'house' darks under the controlled conditions you describe?

One more: when you say to use the cover end cap, would it be better to completely cover the end i.e. put a hat/bag or something over it too, if the idea is to identify light leaks elsewhere in the chain?

Edited April 3, 2022 by BrendanC

[tomato]

Here are the LRGB channels calibrated and stacked in APP, stretched but no gradient removal tool applied, these were registered and normalised in APP then taken into Startools. I had to push the blue channel and I have clipped the data in trying to get a more even background which I think was the original issue? 

Red

Green

Blue

 

[vlaiv]

49 minutes ago, BrendanC said:

I'm not sure what you mean by subtracting one from the other. If I were to open them in, say, Photoshop layers, does this mean I would use the Difference or Exclusion blending mode for example?

Also, given that I'm sure the darks I took in the garden are good - I mean, the scope was swathed in towels, including the camera and focus tube, the cap was on, there was a hat over the cap, and over the primary end, under the towels, all in astro darkness - then I should be fine classing them as the 'house' versions, right? And then I can do the thing you suggest, of taking equivalent darks but with just the cap on, and use them to run this analysis? Or to be on the safe side, should I retake some 'house' darks under the controlled conditions you describe?

One more: when you say to use the cover end cap, would it be better to completely cover the end i.e. put a hat/bag or something over it too, if the idea is to identify light leaks elsewhere in the chain?

one set of darks should be in the same conditions you have when shooting lights - with only front of the scope covered by cap.

another set of darks should really be taken in different conditions. Sometimes IR can cause light leak, and no matter how much "blankets" you put - IR can get thru (it will go thru plastic as well - metal or aluminum foil stops IR).

In order to remove anything scope related, it is best to really do it in house. Put camera cap on, place it on the desk face down or alternatively do this:

I personally don't use aluminum foil - but I do place camera "sensor down" on the desk so that desk acts as additional IR block.

Turn off lights in the room and shoot darks. Simple as that.

 

[BrendanC]

... and THAT'S why people wear aluminium foil hats, right?

Thank you again @vlaiv, I will be doing this exact test over the next night or so.

In the meantime, with slightly more aggressive wiping than I would like, and a possibly clipped background, here's where I'm up to. 

[vlaiv]

1 minute ago, BrendanC said:

... and THAT'S why people wear aluminium foil hats, right?

Sure, to stop unwanted electromagnetic interference with their "sensitive bits"

[BrendanC]

1 hour ago, tomato said:

Here are the LRGB channels calibrated and stacked in APP, stretched but no gradient removal tool applied, these were registered and normalised in APP then taken into Startools. I had to push the blue channel and I have clipped the data in trying to get a more even background which I think was the original issue?

That looks great - thank you. I'm starting to get the hang of the data in StarTools too, I think, but I agree that the background is what's causing me grief - I had to be aggressive with the gradient wipe, much more so than I anticipated. But you know what it's like when you have new kit? When you start, and it's not right, it could be ANYTHING, so I'm really grateful to you for going through the data. I really like your attempt and I may have overcooked mine a bit, but it does seem that I have a direction to follow now: stretching less, wiping more, and trying to identify where that gradient might be coming from.

I only use APP for stacking but I'm increasingly interested in its gradient removal tool, assuming that it doesn't disrupt the linearity of the data too much for StarTools. If only it had a user guide!

Again, thank you everyone for your help. If this is starting to sound like an Oscar acceptance speech then don't worry, I'm not going to slap anyone...

Edited April 3, 2022 by BrendanC

[tomato]

It’s weird but sometimes StarTools does a great job with my data and other times I can’t get a decent result. On your data I thought APP would do a better job but the initial RGB combination was a mile off on colour. I will give it another go in APP, the remove light pollution tool is excellent.

[BrendanC]

I only quite recently learned the star sampling technique in ST for getting the colour right, and it really worked with this one. I know what you mean though, ST can be quite unpredictable.

[tomato]

As promised, another version combining LRGB in APP, further processing in APP, Pixinsight (on the stretched file) and Affinity Photo to balance the colour.

It's just my opinion, but I would be chuffed if I had captured this data.

 

[BrendanC]

Outstanding. I love that you've kept the background from clipping and the stars look amazing.

Thank you so much again. What a relief! I'm going to get more Bodes data to build on this, now that I think I'm doing things right. I'll probably go down from 60s subs to 45s luminance, and 180s to 120s for the RGB, just to see the effect it has.

I'm still puzzled by the amount of vignetting, but I'll keep seeing what I can do about it. It could just be down to my shooting conditions and the camera being much more sensitive than the DSLR.

Cheers, Brendan

[tomato]

Thanks, it's quite something for me to be complemented on my processing, its an area where I still have a long way to go.

I think taking shorter subs is worth a try, but if you upload another set to have a go with, I'll have to beef up my router!😄