Looking for advice on image processing - Forums

Topic starter

Dec 9, 2021 · Edited: Dec 9, 2021

The following image was about 3 1/2 hrs of data on the Crescent nebula. I know the stars are bloated, it's noisy and the colours don't seem right, so just wondering what any of you would do if you saw this kind of result (apart from cry). Clearly more integration time would help, plus better guiding (this was 1.5 arc seconds typically). I'm really new to nebulas (this is my first one), so any sage advice would be appreciated. BTW: I would love lots of criticism as that is the only way I'm going to learn.

Quick and dirty Crescent

Some things to note:

* First time I've used mono camera, so the whole mono + filter stuff is new to me
* First time using Pixinsight so have relied on scripts too much I know
* My mount (HEQ5 Pro) is at its payload limits with my 200PDS+ASI1600MM+8xfilter wheel+guide scope+coma corrector. I have belt-modded it though.
* I'm using an LED flat panel on its lowest setting for flats and using dark flats rather than bias
* I really struggle to get rid of DEC backlash and the RA results seem quite poor in PHD2
* PHD2 says that I'm guiding to within about 1/10th pixel on my guide camera

Helpful Respectful Engaging

Chris Bailey

Dec 9, 2021

If it were me I would separate the RGB elements from the Narrowband ones so -

take the RGB stack. Stretch. StarNet with star mask ticked and keep only the stars. boost the saturation of the stars to taste, run morphology erosion with 3 rounds at 0.15 to knock back the stars a little.

take the Ha and Oiii stack and make a synth green in pixel math so r=ha, g =0.5*ha+0.5 Oiii, b= Oiii. Stretch this to a similar background levels as the RGB. StarNet and throw away the stars.

now in pixelmath add the two together I.e RGB stars + NB Nebula.

Helpful Insightful

andrea tasselli

Frequent contributor

Dec 9, 2021

It isn't bad at all. Just keep at it. Use the NBRGB script to get the best of both NB and RGB/LRGB imaging.

Andre Vilhena

Dec 9, 2021

Hello Andy,

In terms of guiding you seem to be talking about an horror movie but actually looking at the stars, they seem quite round across the FOV. Guiding is a tricky thing but in the end of tha day, the criteria should be how the stars look and yours look fairly good.
In terms of processing, there is so much you can do with just a couple of hours. It's noisy, of course, but if you kill all the noise you'll have a plasticized photo. The stars are indeed a bit bloated but not that much. Some are overly colour satured and I would try to avoid that.
Nebula is fairly ok - again, with a couple of hours you can't expect miracles. Still, you can start seeing the blue outer shell, despite the bluish tone is not there (but it is not easy to get - I have a photo with much longer integration time and needed to use some tricks to highlight the blue tone).
It seems to exist a slight gradient in the diagonal but seems correctable..

But again, I think it's a quite good result for such small integration time and you should he happy with it.

Cheers,
André

Helpful Supportive

Jonathan Piques

Dec 9, 2021

Ok, my first piece of feedback is to shut it: this is an *excellent* first go at both using a mono camera and at using PixInsight especially. My first images looked like a train wreck compared to this. AND you incorporated RGB stars, which ain't easy to do. So seriously, be proud, and I'm not just saying that to be nice: this is a great image.

Now, as for other things:

Actually the biggest thing you can do to improve the image isn't your processing, it's your capture. I want you to get at least 2x and preferably 3x more time on target. 50 min of Oiii ain't gonna cut it, nor is 1.5 hours of Ha. Oiii signal is already almost always weak, particularly with the Crescent, and it's also one of the most interesting parts of the Crescent, as it's what comprises the outer gas envelope. Just humor me and get 6 hours each of Ha and Oiii and then report back. I promise you'll be much happier.
Next, blink your raw images using Blink in PixInsight before you run them through WBPP and integrate them. Throw out any crappy subs.
If you do those two things, I promise you'll see huge gains in the quality of your image and you won't have to use so much noise reduction, if any (particularly in narrowband). Yes, it's a drag to spend more time gathering data, but it pays off and makes your life way easier when it comes to processing.
Consider ditching the RGB stars for now if you're doing narrowband. First learn to process narrowband well and then figure out how to get those RGB stars in there and have them look good. With narrowband, what makes the image is the structure / detail and the color of the nebulosity: getting the stars to be RGB in nature is like a tertiary priority.
On basic narrowband processing steps, try this:

Blink your subs, discard any that are obviously bad
Use WBPP to create integrated masters. Later on, eventually learn to use WBPP to take everything up through registration, then run the NormalizeScaleGradient script, and then integrate.
First thing you do, run the MureDenoise script on your new masters. Yeah, you gotta figure out your detector settings to do so but it's the best denoising routine in PI and it's literally push-button easy to run after you figure those settings out the first time.
Do a basic stretch on each using ScreenTransferFunction and Histogram transformation: after you hit autostretch, click and drag the little bottom left corner of STF to the bottom of HistogramTransformation, and it'll automatically set the settings to do a stretch. When you're doing this first stretch, it's better to be a little dimmer than a little too bright, or you'll lose information / color. No need to get crazy dim, and definitely don't black clip anything, but just don't be too bright--you can adjust the sliders in STF or in HT to adjust. Yes, there are a million other ways to do your stretches but if you're just starting out in Pixinsight this is the easiest way for narrowband. No matter what, try and get both the Ha and the Oiii to be about the same level of brightness (again, to start).
Now that you have non-linear masters, combine them into a color image using Pixelmath. There are a million different formulas for this, and you should always just tinker with it no matter what formula you start with. Some are simple, some are complex. Here's a neat one to try for bi-color like you've got:

Red: iif(ha>0.15,ha,(ha*0.8)+(oiii*0.2))
Green: iif(ha>0.5,1-(1-oiii)*(1-(ha-0.5)),oiii*(ha+0.5))
Blue: iif(oiii>0.1,oiii,(ha*0.3)+(oiii*0.2))

When picking your color combination, don't worry about your star color. We will fix that later. Focus on the nebula.
After you have your color image, run StarXterminator on it so all that's remaining is the pretty stuff. Play with CurvesTransformation (I like the 'c' curve) and ColorSaturation to get your image looking prettier. Don't oversaturate things too much.
When you're satisfied with the above step, apply one of your mono images as luminance using ChannelCombination on Clab mode. Uncheck a and b and just leave L. Denote the image you'd like to use as luminance in the little box and then hit apply. As for which image you use for luminance, sometimes you can just get away with Ha, as it's usually the strongest signal. For the Crescent, if you do this, you'll miss out on all that pretty Oiii structure. So maybe create a synthetic luminance by either taking the average of your Ha and Oiii masters in pixelmath, or extracting luminance from your original color combination (look for a little button on the top left of your toolbar to extract luminance from a color image) before you ran StarXterminator.
When you apply a luminance image with stars in it to a starless color image, it has the effect of basically making all the stars white. Some people care about this. I generally don't. Again, what makes the image most impactful is the nebula: as long as your stars aren't some weird distracting color you're good. You can learn to put RGB stars in later. It's not hard, just another darn step.
Now you're basically done. From here, just make curves adjustments to taste. Try and avoid noise reduction if you can, and if you have to do it, use a light touch.

Well I just wrote a hell of a lot more than I planned to. Sorry for the novel, but hope it helps some. And again, great first image: you should be proud. Try getting way way more integration time and I guarantee you'll be happier.

Helpful Insightful Engaging Supportive

Andy Wray

Topic starter

Dec 9, 2021

Jonathan Piques:
Well I just wrote a hell of a lot more than I planned to. Sorry for the novel, but hope it helps some.

That helps a lot and makes a load of sense. Thank you

Well Written Respectful

Andy Wray

Topic starter

Dec 9, 2021

Andre Vilhena:
Guiding is a tricky thing but in the end of tha day, the criteria should be how the stars look and yours look fairly good.

Thanks for that re-assurance ... I'll stop stressing about the guidance so much right now and focus on getting more integration time. Appreciate the feedback.

Well Written Respectful

Andre Vilhena

Dec 9, 2021

Andy Wray:
Andre Vilhena:
Guiding is a tricky thing but in the end of tha day, the criteria should be how the stars look and yours look fairly good.

Thanks for that re-assurance ... I'll stop stressing about the guidance so much right now and focus on getting more integration time. Appreciate the feedback.

*Andy, you're welcome and be happy with your photo, it's really good for such short integration time. Recently I'm not taking photos with less than 16 hours integration and I can tell you they're so much easier to process and the level of detail is much higher - this would be my biggest advice.
Let me know if I can be of any further help.
Cheers.

Well Written Insightful Respectful Concise Supportive

Andy Wray

Topic starter

Dec 10, 2021

Andre Vilhena:
Recently I'm not taking photos with less than 16 hours integration and I can tell you they're so much easier to process and the level of detail is much higher

That makes loads of sense ... I've only had 6 hours of clear skies in one and a half months where I live, but hopefully the skies will clear later in the month and into the new year. It's definitely not a hobby for the impatient.

Well Written

Mau_Bard

Dec 11, 2021

Hi Andy, you can be more than satisfied with the result you got, seriously. This is a target that requires longer exposures to get the best out of it, nevertheless I find you did well. Not to mention the complexities of PixInsight, that you, again, tamed well for a first run!
A ton of extremely good advice has been given here by Chris, Andrea, Jonathan and Andre.
Let me only stress that, according to my experience in astrophoto, it is preferrable to try a new thing (HW, SW or processing) at a time, in order to isolate issues, and, even more important, to find your way to use the newly added feature in a creative way.
As @Jonathan Piques already stated, I would for instance concentrate on narrowband Ha+OIII and make some practice there, in order to settle on a nice and personal nebula color balance process, before embarking in more complex processing.

CS, Mau

Helpful Respectful Supportive