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:

Honeyview_60s.jpg (155.1 KiB) Viewed 2585 times

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 thought it was significant that the glows are red, green and blue (and probably luminance bottom right) so wondered if it was my flats. So, I tried just stacking and processing ten minutes of each channel with no calibration at all, but that still exhibits colours glowing around the edges:

no calibs.jpg (157.95 KiB) Viewed 2585 times

Any ideas or pointers for what could be going wrong here?

Thanks, Brendan

[BrendanC]

So I've been looking into this and frankly, I'm stumped. This is worrying. I've been through everything and I can't figure out what I'm doing wrong.

Would anyone have the time and/or inclination to take a look at my masters, and ten each of my LRGB subs, and let me know if there's anything glaringly wrong?

If so, they're here: ten each of LRGB 60s, plus masters: https://1drv.ms/u/s!AqovBuVZMwj3kuUA4E9 ... A?e=Wbv7Cu - masters produced in APP.

Thanks, Brendan

[happy-kat]

Assuming any edge artefacts were copped out, I would in Wipe mask the galaxy.

[BrendanC]

I heavily cropped around the edges, and I don't think the Wipe module could fix this. It's a real problem somewhere along the line, between acquisition, calibration and processing. I don't know if it's hardware, software, filters, camera, whatever. I'm stumped. And a bit worried.

[Mike in Rancho]

Brendan I tried but I don't think the OneDrive thing is working out very well for me. Almost dial-up slow it seemed, but even so both (long) attempts ended up with a "failed download" message.

Google Drive or Dropbox might be appreciably faster. From what I can tell you'll have about 2GB of data there, which is big but should be doable. I've downloaded plenty of zips that big before.

Kind of too many to do individually, though I suppose that's possible. Or maybe I can try to do each folder/subfolder one by one.

What is the telescope?

[Mike in Rancho]

Hi Brendan,

I was bored so I downloaded them all in small chunks, folder by folder, and that worked a bit better. Though still slowly.

I didn't get much out of it, but I think only 10 minutes total per filter was provided? And I had to sledgehammer it with Wipe too, adding to the lack of galaxy.

Every program does its own thing I guess, but your calibration masters were 32-bit, and DSS choked on that. So, ASTAP to the rescue. Got each filter stacked there and then registered those results in DSS. Composited in ST. Major flats correction/vignetting issues. Could not get a flat field in Wipe for both color and luminance without using vignetting settings, DAF4, and something like 98% aggressiveness. And even then I had to use Contrast to squeeze the shadows along with SS-Isolate. I didn't run any other enhancement modules here for this test.

A couple things I noticed:

Same exposure for L and each color filter. I wouldn't expect that to be the desired case.

The levels of the dark flat masters varied wildly by filter, at least according to ASTAP and Fitsworks. Again I don't know APP, but that seems abnormal to me. Flats? Sure, not that big a deal perhaps. But dark flats should be pretty much like...darks. In fact, since they would not be 60s like the lights, but instead timed to the flats, they should actually be darker than darks. And yet they aren't.

So, maybe look into your APP calibration settings to see what was created here as far as calibration masters. And perhaps review your underlying dark flat subs, and compare them to your darks subs.

Maybe also check acquisition. Did you use the same software to capture flats and dark flats that you did for lights and darks? Sometimes different programs will use different camera drivers (as in native vs ASCOM), and that can very much affect your results.

Anyway that's my guess - miscalibration.

As to your no-calibration image, there's still obvious vignetting and I would imagine you'd get colors in your vignetting. These are discrete RGB filters, after all, and you've composited them. Also I noted that two of the filters (I forget which) were flipped 180° from the other two, and that might be adding to the rainbow effect here.

Here's my ST result from your 10m per filter samples. Not much there huh? If you want to link your full 9 hour stack I can probably erase the galaxy from that too!

BrendanC NGC4236 sample ST8 1A.jpg (310.27 KiB) Viewed 2543 times

[BrendanC]

Wow. Thank you so much Mike, this is incredibly helpful.

I'll try and go through your points but I'm finding the quote feature irksome so I'm just going to use italics...

I was bored so I downloaded them all in small chunks, folder by folder, and that worked a bit better. Though still slowly. - no idea about this, blame OneDrive!

I didn't get much out of it, but I think only 10 minutes total per filter was provided? - that's correct, I only provided 10 subs to avoid huge amounts of download, just to give an idea of the subs and masters I was working with

Same exposure for L and each color filter. I wouldn't expect that to be the desired case. - so this is where my newbness comes in with LRGB. I thought they would all have the same exposure time, but I'm learning that's not the case.

The levels of the dark flat masters varied wildly by filter, at least according to ASTAP and Fitsworks. Again I don't know APP, but that seems abnormal to me. - me too, I cannot account for this.

dark flats should be pretty much like...darks. In fact, since they would not be 60s like the lights, but instead timed to the flats, they should actually be darker than darks. - I don't understand this either. When the darks were coming through they seemed very, well, light! So are you saying the darks are wrong? They were taken with the camera shut in the fridge, with a scarf around it, to shut out all the light.

So, maybe look into your APP calibration settings to see what was created here as far as calibration masters. And perhaps review your underlying dark flat subs, and compare them to your darks subs. - yep, I think I need to totally revisit calibration

Did you use the same software to capture flats and dark flats that you did for lights and darks? Yes - all via APT, using its flats tool to calculate the ideal exposure time, via the native ZWO drivers.

Also I noted that two of the filters (I forget which) were flipped 180° from the other two, and that might be adding to the rainbow effect here. - totally do not understand this! Do you mean that the images were flipped? In which case, is this where shooting before/after the meridian flip comes in? As in, assuming there is vignetting with the filters, and I take my L and R before the flip, and my G and B after it, could that have caused this? If so then that's another learning, and would tie in with what I immediately spotted, that the L, R, G and B were all kind of around the corners of the image.

If you want to link your full 9 hour stack I can probably erase the galaxy from that too! - well, here's a link to the four stacked LRGB files if you'd like to take a look: https://1drv.ms/u/s!AqovBuVZMwj3kuU6Le1 ... A?e=1KcI1i The amount of exposure time for each one is in the filename, so for example L 12720s.fits is the Luminance and has 12,720s of exposure. The reason the values are different is because I had some glitches during the shoot, and the Compose module should be able to balance these out if you specify the amounts there (which I'm entirely sure you know about!)

So, in summary:
* It would be great if you could take a quick look at the four fully stacked channels as per that link
* By how much more do my L subs need to be, than my RGB subs? Or is that something I should just look up myself? I'm fairly sure that when I checked Astrobin, people were using equal durations, but that could have just me not looking hard enough.
* Do you think my dark flats might benefit from a longer exposure time? I'm using an LED panel with a perspex sheet on it, and calculated the flats exposure time using APT's flats tool, and just used the same settings for the darks. From what I'm reading, it seems perhaps ZWO cameras prefer longer flats, so I'm considering retaking them, with some paper added to the panel to darken it, and get longer flats.
* Am I right about needing to take equal filter exposure pre- and post-flip?

That's all (for now). I am incredibly grateful for your help Mike.

Regards
Brendan

[Mike in Rancho]

Hi Brendan,

I'll just do things without quotes as well, and also not in order (since it is getting pretty busy). So...stream of consciousness.

Don't worry about the meridian flip. Your registration will get everything lined up in the end. The only reason I mentioned it was as a possible explanation for the sample you provided with "no calibration." In addition to the uncorrected vignetting, you've also got what I presume is LP gradient. And with those things flipped relative to each other for 2 of the sets, well, it seems a reasonable explanation for how you'd end up with remnant colors in all sorts of places after composition. Note also that in addition to the flip, there's a bit of a rotation difference for a couple filters as well. Again, don't worry about this.

Can you provide an upload of your individual calibration subs (1 each/filter is good enough) - dark, flat, and dark flat? I think this may help more than the already-generated masters.

Did you say you took both darks and dark flats wrapped in a towel in the fridge?

As for exposure times, there are two here of course. One is integration for each filter, and yes it can be adjusted for in ST compose (for purposes of any Synth L, at least). A possible rule of thumb is to acquire L and (total) RGB equally. Thus, if you ended up with an hour of L, you would take 20m each of R, G, and B.

The other, which I was talking about above, is subexposure times for each filter. You are using 60s for everything. This may be okay, but keep in mind that L is gathering more light than any of RGB. The key is to make sure you are exposed enough to get out of the read noise, balanced against how much you are oversaturating bright things like stars and how many files you want to store/stack.

So, does APT have any kind of subexposure helper or histogram/pixel counts to let you know if a sub is good? My guess is that you will be able to go a bit longer on the RGB than the L to reach whatever appropriate compromise you want for exposures levels and blown pixels.

But right now that's just making adjustments to fine tune your lights acquisition. Probably more important to figure out the calibration matter first.

When I get a chance later I'll try out your longer LRGB stacks and see what ST and Wipe can do with them. But I think looking at the calibration subframes and looking a bit more into dark flats may reveal more for the investigation here. I did open each filter stack individually in ST for an AutoDev look-see, and they showed the vignetting and gradient that seem to be going on here. Gradient is fine. Vignetting issue and flats calibration needs to be figured out.

[BrendanC]

OK, so first up, here's the link to the individual calibration subs, as requested: https://1drv.ms/u/s!AqovBuVZMwj3kucbF1c ... w?e=vwbD6T

I've already done some work on flats and dark flats, and have new versions that have 2+ seconds' exposure time. I know the flats might not be quite properly oriented, but I'm giving them a shot with APP restacking everything as I type.

Thanks for telling me I don't need to worry about the flip and the LP gradient!

I did take darks and dark flats wrapped in a towel in the fridge - the camera, that is, not me. The only reason for that is simply because the fridge struck me as a good dark area. I'd tried taking darks with the scope in a cupboard swathed in blankets, but couldn't be sure light wasn't getting in. Having said that, the dark flats I just took were non-fridge versions. Fingers crossed on that point.

I like that rule of thumb for total exposure. Works for me.

As for individual subs, well APT's histogram is not great, and this is something I've asked about on their forum. The DSLR version is really handy, and I used it to very quickly zero in on what worked. But the CCD version just has a spike at either end and nothing in the middle. I've been told I should have the spike to the left away from the edge, and as little a spike as possible on the right to stop stars getting overexposed, but I can barely tell a difference between one exposure duration and another. This is why I've been relying on tables such as this one to help get a starting point: https://www.cloudynights.com/topic/5738 ... be-qhy163/ - but perhaps this is another thing I need to look into more.

Anyway, thank you again for helping out a total noobtard.

Cheers, Brendan

[Mike in Rancho]

Hi Brendan. Cool thanks for putting up the subs.

Not sure I did anything more than confuse myself though!

The uploaded files look good to me. More or less everything falls where it ought to. The flats are maybe a little low, with an ADU median around 10K, but I don't know if that's fatal to anything. The dark sub has a median of 336, and the dark flats either 304 or 320, which per above is what I would expect. Well, the dark flats had a temp set of -5C compared to the -15C of all the others, but I don't know if that's enough to throw things off?

And I also don't understand the APP masters, but that could just be program idiosyncrasies. Don't use it, so I have nothing to compare. You might double check that all your subs (flats, dark flats) used in creation of the masters are matching. Like, you didn't leave something in the folder with different gain, or offset, or accidentally include a flat in the dark flat folder. Dunno.

Yes there are a lot of threads on CN about the 1600 being finicky. Some say use bias, some say dark flats. Some say expose for longer than 1 second. I guess just experiment and keep researching, try to match what has worked well for others.

For a reality check you might just try out ASTAP too and see how it compares to APP. In the lists, after you click on Analyze, you'll see the background ADU levels. Might be able to quickly flag if you have a wrong sub in some of your folders (even if you don't bother going through stacking in ASTAP).

Another thing I just noticed in your full stacks, as compared to the 10 file samples from earlier, is just how much field rotation you have going on even within filter sessions. That's going to cause some confusion as to whether we are really looking at a vignetting correction (flats) problem, or you just have your gradients being rotated (and flipped) around so that it looks like vignetting, if that makes sense.

It also means significant cropping will be required, especially after all 4 stacks are composited. I may not have done enough of that here and perhaps I'll try again later with even more crop.

It still took a strong Wipe, high aggressiveness, artificial flats, and a big DAF. I then think I squashed the background so much that I didn't get a very good SVD. And then used an SS Isolate followed by SS brighten to bring the target back into better view. Best I could do quickly but probably not as good as you'd like to get, detail wise, for your 9 hours integration.

Well, I should probably get to work.

BrendanC NGC4236 allstacks ST8 1A.jpg (491.15 KiB) Viewed 2522 times