First LRGB image - strange RGB glows around edges

[BrendanC]

Hi all,

I've recently moved from DSLR to mono imaging. I'm getting the hang of narrowband, but my first LRGB image is something of  a disappointment.

It's NGC 4236 which I've captured a bit of before, and love its blue disc, so I thought it would be a good first LRGB target to compare with.

It came out like this:

This is about 9 hours of 60s exposures, at gain 75, offset 21, ASI1600MM-Cool, ZWO LRGB 1.25 inch filters, with 25 flats, 25 dark flats and 50 darks.

I think it's significant that I have red, green and blue glows, which implies something is wrong with my calibration, perhaps my flats? 

Any takers?

Thanks, Brendan

[Varavall]

Have you tried stacking the images without using the flats to see if that Xmas tree affect persists?

[BrendanC]

Good call - and this is what I get from just stacking 10 mins of each channel without any calibration files - still a decided RGB glow around the edges. So, probably not calibration then.

Any ideas what could be causing this? I'm a total newb with mono imaging.

[kirkster501]

This could still be caused by lack of calibration or reflections within the tube.  You need to calibrate the lights first.

[Varavall]

What OTA are you using; are you in a light polluted zone; was this a very dewy night, was the Saharan dust passing over; how are you illuminating and creating the flats? Anything else you can think of that could help diagnosis would be good!

[Varavall]

PS: You are creating flats for each filter?

[BrendanC]

Thanks for the questions! It's all very useful.

I am calibrating the flats with flat darks, taken at the same time exposure time/gain/offset as the flats.

I am also using darks matched to the same exposure time as the lights.

The flats were taken separately for each filter, using the flats tool in APT, which I also used to create the dark flats.

I use a 130PDS in a Bortle 4 zone, it wasn't a dewy night, there was no Moon, no dust. The object was quite low in the sky, which means it could have picked up some light pollution from streetlamps I guess. But I would have expected this to be a constant gradient, not separate colours in each corner.

I take the flats using an LED screen like this, with a perspex cover on it: https://www.amazon.co.uk/Stone-TH-Drawing-Lighting-Animation/dp/B07N128B74/ref=sr_1_7?crid=34ZCCBXXBOH1G&keywords=led+tracing+pad+stone&qid=1648571555&sprefix=led+tracing+pad+stone%2Caps%2C64&sr=8-7 - which worked fine for two years with the DSLR. Perhaps this is the problem? It's not an especially sophisticated approach but it worked fine with narrowband with this camera, and worked fine with the DSLR.

One thing: I am using the same darks across all channels. This is correct, right? The flats and flat darks are per channel, but the darks can be used across all?

Perhaps I should try just calibrating the lights with the flat darks to see what happens, then calibrating just with the darks?

Is there anything else I could be looking into?

Edited March 29, 2022 by BrendanC

[Elp]

Was it all shot on the same night with the cam/scope in the same orientation across all filter changes? I find with LRGB, especially with the moon out, I get some kind of gradient in the image mostly from terrestrial light pollution.

[Elp]

Are the peaks of your RGB also aligned, from the centre of your image id guess they are.

[symmetal]

Brendan,

Are you using Startools for processing? This is a typical effect of not cropping off the stacking artifacts from around the edge of the frame, before using the 'Wipe' tool.

Alan

[BrendanC]

1 hour ago, Elp said:

Was it all shot on the same night with the cam/scope in the same orientation across all filter changes? I find with LRGB, especially with the moon out, I get some kind of gradient in the image mostly from terrestrial light pollution.

Two nights, but under virtually identical conditions, and no Moon.

 

1 hour ago, Elp said:

Are the peaks of your RGB also aligned, from the centre of your image id guess they are.

I don't know! How would I check this?

 

48 minutes ago, symmetal said:

Brendan,

Are you using Startools for processing? This is a typical effect of not cropping off the stacking artifacts from around the edge of the frame, before using the 'Wipe' tool.

Alan

I am, and this is a very good point, but I have definitely made sure I'm cropping out the artefacts.

I'm doing another run with some different calibration frames so let's see what transpires.

Thank you so much for the responses.

[Elp]

To align RGB peaks, whatever editing software youre using if it's half decent will have an option to display the histogram, in PS it's in the top RH pane in default layout, you select the histogram tab and click on the small triangle to select the expanded view option which will display r, g and b in separate graphs. You then adjust the curve levels for each channel until the peaks of each channel are roughly aligned. Personally though I don't believe this is your issue. 

As you have been doing ensure you take the flats in the same conditions as you've taken the lights for each filter, obviously adjusting the exposure time in order to take the flats.

[BrendanC]

Thanks. I'm using StarTools so I expect that'll be looking after levels that aren't visible to me (I also use Photoshop, but the ASI1600 produces FITS files which I'd have to convert to TIFF to view them, or I can use FITS Liberator).

I just thought I might have been using the wrong flats and flat darks, but I get something similar (and yes, there does seem to be some walking noise but this is with very few subs and I might just increase the dithering rate to handle this).

I'm stumped, frankly. This is worrying.

Edited March 29, 2022 by BrendanC

[Elp]

Also try a slightly higher gain just above unity to reduce the read noise. Not too high as theres a limited advantage beyond a certain point (see the graphs on zwos product pages). There's a nice range of colour in the stars in the image though which is nice.

The issue kind of looks like how the 294mc is difficult to create flats for, but as you're shooting mono with a different camera...

Your filters do cover the sensor fully (would have thought so as theres no clipping to the edges).

Edited March 29, 2022 by Elp

[BrendanC]

Ideally I wouldn't have to retake all nine hours' worth! They were shot with gain 75 which should be fine, from what I've read. The noise really isn't the issue, I can fix that (I haven't applied any denoise whatsoever to these images).

I hope the filters cover the sensors because they're by the same manufacturer and came in a bundle. I have read that there may be some vignetting with the 1.25 inch mounted filters but this is far beyond what I would have expected.

I can believe that the ASI1600 is difficult to create flats for, but I've been through the APT flats tool which should be creating the ideal exposure times for each flat. 

All I can think is that I need to redo the flats and take it from there.

The other possibility is: would anyone be willing to take a look at some of my raw LRGB subs, master flat, master dark and master dark flat, and let me know what they think?

If so, they're here: ten each of LRGB, plus masters: https://1drv.ms/u/s!AqovBuVZMwj3kuUA4E9LynWaVjMvQA?e=Wbv7Cu - masters produced in APP.

Thanks, Brendan

[Varavall]

3 hours ago, BrendanC said:

I take the flats using an LED screen like this

I have the same camera and I use the very same illuminated screen for flats, but I use NINA for the flats wizard and I have never seen the problem you have. I use a SW 20Opds and I did have a problem of light entering around the primary mirror so I made a black rubber shower cap to cover the end of the OTA; problem solved. I believe there was some sort of issue with the older ZWO filters, but I have only used Astronomik and Astrodon filters. Are the filters the correct way round? Any possibility of a light leak around the filer wheel connections?

[BrendanC]

OK, thanks, so it's not the screen, or the camera.

The filters are 1.25 inch mounted, so I don't see how I could have put them on the wrong way around, as they screw in.

I have no idea if there's a possibility of a leak around the filter wheel connections. I also have a cap around the end of my 130PDS to stop light entering the primary.

If there's a problem with the old filters then I don't see how they're actually worth the money if this is what they produce! Surely I'm doing something wrong here.

I haven't given NINA a go, but I might, and see if the flats wizard works any better with that. Thanks for the suggestion.

This is rapidly turning into one of those threads where you read it and shiver and think 'I'm so glad that isn't me', except that this time, it's me.

 

Edited March 29, 2022 by BrendanC

[Varavall]

10 minutes ago, BrendanC said:

so I don't see how I could have put them on the wrong way around, as they screw in.

The filter wheel itself can be attached in 2 ways. The camera should be attached to the side opposite the removable cover that gives access to the wheel so that the filters are screwed in towards the camera, if that makes sense.

[BrendanC]

Again, check: that's how I have the camera attached, with the EFW lettering on the camera side, the screw holes on the other removable side. Good thinking though.

I just processed some of the individual subs in StarTools as an LRGB composition, and they look exactly as I'd imagine - a clear vignette, which the flats would calibrate out, but none of this extraneous colour.

So I'm starting to think the problem is as I first suspected, something to do with the flats.

Perhaps there was a leak when I took them. They were taken in twilight, with the LED screen pressed up against the tube, which has always been OK in the past with the DSLR. But perhaps with the additional sensitivity of this camera, I needed to be more rigorous. So tomorrow I'm going to put the whole OTA, camera, filters and all, into a very dark room, covered with a towel or two, retake the flats, and dark flats, and pray.

 

Edited March 29, 2022 by BrendanC

[Varavall]

Good plan. I await the result with interest.

[Elp]

Don't know how you're gonna get the imaging train exactly the same as when you took the lights though...

Try taking the flats manually with your own exposure setting, just aim for the histogram peaks to be around halfway across the graph, it doesn't have to be exact as I've used flats with peaks one third way across too.

[BrendanC]

I probably won't be able to get it exactly right, but I know I had the camera at my 'neutral' position where it's aligned with the scope (which is why the diff spikes are nice and horizontal /vertical) so I should be able to get close. 

It seems ZWO cameras prefer longer flats, so I'm going to add some paper to my light pad when using the flats tool in APT. If that doesn't work I'm going to download NINA and try the flats wizard. 

Edited March 29, 2022 by BrendanC

[symmetal]

Brendan,

I stacked your files in Astro Pixel Processor and loaded the results into Startools. Flats have worked well. I've cropped a lot less than you have as shown. I couldn't get a useable stretch with AutoDev as the image is very noisy and the selected area on the DSO very small, so I used FilmDev for a quick stretch until the background noise became too noticeable.

As a result you can see that the DSO is itself quite dim but there are no coloured patches. Your posted images are significantly overstretched by AutoDev so any slight colour errors are being exaggerated though not sure why they are as localised as in your first image.

Startools 'Crop' selection just to avoid any stacking artifacts. No vignetting or dust spots so your flats work fine.

 

'Wipe' Luminance result is good. Bright star top left causing some flaring.

'Wipe' RGB result is good too with no significant coloured patches.

Result from 'Colour' module after 'FilmDev' stretch. I could have stretched it a bit more seeing the final result, but the background noise would start to dominate. 

You need longer RGB exposures and many more of them too including Luminance. At a minimum I go for 60 Luminance of 60s, and 20 each of RG and B of 180s, giving 1 hour total each of LRG and B. Then add more luminance exposures, (and sometimes more RG and B ) until the background noise is acceptable. 

I made this chart many years ago when the 1600 came out to get optimum exposure values so that the camera read noise is swamped by the sky background and so doesn't contribute to the final result to any significant degree. The 10 x RN^2 column is the one I use  for a suitable sky background ADU value.

I don't know what gain you used but when starting out, unity gain is recommended. When it first came out using variable offset with gain was suggested, used to maximize the dynamic range but this is a real pain and it was generally decided to use a fixed offset to cope with any gain setting but avoiding black clipping in the image. 50 was the generally accepted value though I used 64 in the end as 50 still gave some black clipping in the images. Also for the 1600 use flat and dark flat exposures of at least 2 seconds or more to avoid the different camera internal dark processing for short exposures, meaning short flats and dark flats of a second or less will have the wrong noise profile to calibrate your long exposure images correctly. This means having many sheets of paper in front of the scope for flats to get long exposures.

Your Luminance subs have a median ADU (sky background level) of 1800 ADU which seems reasonable, but your R,G and B subs median ADUs are 608, 624, and 528 which are much too low. As a rule your RGB exposures should be 3 times the duration of your Luminance exposures to achieve the same sky background ADU level and so swamp the read noise the same.

Using an offset of 50, 800ADU is added by the offset alone (50 x 16, converting 12bit to 16 bit) so any image taken should be around 750 to 800ADU or so at a minimum, depending on its internal ADC (Analogue to Digital Converter) setup. Let me know what gain and offset you're using and are your RGB gain/offset settings different to L, so giving such low image background ADU values. My 1600 flat darks subs have a median ADU of around 860 ADU at offset 50.

Hope this helps. 🙂

Alan

Edited March 30, 2022 by symmetal

[ONIKKINEN]

There are several bright stars within a few degrees of NGC 4236 and some of these can be in a sweet spot where they just about hit the edge of your coma corrector and cause reflections off the lenses, at least thats what i think happened for me (i am also imaging the galaxy).

See the rainbow reflection thingy in the bottom left? That happens when i have a bright star at just the right distance from my image center.

Also had the same happen on a few other targets so its not unique to this target, but i do find that the range of distances is quite narrow. Just a bit closer or further and this goes away, but it can actually be worse than it is in my shot if i get just a bit further away from the bright stars.

Now it doesn't really look the same to me, yours looks more like a flats/gradient removal issue rather than a well defined halo like mine, but something like this could play a part as well.

[BrendanC]

Wow. Thank you so much everyone.

First up, @ONIKKINEN, that's very interesting to note. I'm wondering whether this was a bad target for my first LRGB. It's very low in my sky, which could be introducing some light pollution, and if what you say is true about neighbouring bright objects, well that could also be contributing to my issues here.

But to go on to @symmetal - thank you, thank you, thank you. This is amazing, to think you spent all that time to look into this for me and share your findings.

Just to go through your points:

• I'm glad the flats look fine, but I'm still going to give them another go today. From what I've read, and as you say, it seems ZWO cameras might not like very short flats, so I'm going to add some paper on top of my LED pad to try and give them more time, and see what happens.
• I cropped very far in simply because I wanted to be sure of no artefacts. Also I was running a macro that helped me process more quickly, over and over again, and I just had that crop factor in the macro.
• I understand why you used FilmDev because of the noise, but I'd very much prefer to be using AutoDev because that's a sign of good data.
• When you say the Wipe result is good with RGB, it looks pretty bad to me! Those coloured patches are exactly what is causing me grief.
• I'm in a way not surprised that the image is noisy because I only shared ten subs of each channel, at 60s each, to avoid inundating you with a huge download, the total amount of which is around 20GB! In total I took 9 hours of 60s exposures, at gain 75, offset 21, figures I landed on after doing research into what other people are using, and also this page: https://www.cloudynights.com/topic/573886-sub-exposure-tables-for-asi-1600-and-maybe-qhy163/
• When you say my L is fine but the ADU for R, G and B is too low, I had no idea L needed to be three times the duration. When I look on Astrobin, people seem to generally be using the same exposure times across all channels?
• Finally, while I've got you(!), would it be possible please to take a look at the fully stacked, calibrated LRGB subs in StarTools for me, and tell me what you think? I'm wondering whether you're getting the full picture (literally) by me just sharing ten subs. If so, they're here: https://1drv.ms/u/s!AqovBuVZMwj3kuU6Le1hAw_U3tQubA?e=8G6V4W. These were stacked and calibrated in APP. It would just be interesting to see what you make of them. The filenames have the total amount of exposures time in them eg L 12720s.fits is the Luminance and is 12,720s of exposure. The times are a bit different for each channel simply because a couple of things went wrong during the shoot. These can be compensated for in the StarTools Compose module by specifying the exposure times there.

So in summary I need to:

• Retake my flats and dark flats at longer than 2 seconds
• Go back to unity gain and offset 50 (for now, bearing in mind you went up to 64 eventually)
• Think about shorter L subs and.or RGB subs (although I don't know how I'll do this without overexposing the RGB?)
• Learn much more about ADUs! Coming from DSLR land I'm all clued up with ISOs etc. One problem I have with APT is its histogram, which doesn't really tell me what I need to know, I think. I might give NINA a try and see what that says.
• One more final question: do the master dark, flat darks and flats look ok to you? The darks don't seem very, well, dark to me! They were taken with the camera wrapped in a scarf, in the fridge, all at -15C, just to make sure there was no light coming in.

Does this make sense?

Thanks (again)
Brendan

Edited March 30, 2022 by BrendanC