Iris Nebula - Where did I go wrong ?

[Rico]

My setup:

• Telescope: Skywatcher ED80
• Camera:    Modified 700D
• Guiding:    QHY5L-ii m + finderscope
• Image Capture: Astrophotography tool
• Processing: DSS + PS or Pixinsight
• Filters:       Astronomik CLS clip on

Over the last week I managed to get two nights with the telescope, both of which were spent capturing the iris nebula. On the first night I got about 1.5 hours and on the second night I managed a whopping 7.5 hours. I give thanks to the recently purchased 700d AC power adapter for that long stint. I live in a relatively light polluted area, but I can still manage to see the milky way on a clear night. Now, I usually image relatively easy images ... orions nebula, horsehead nebula, rosette, heart and soul, using the above setup as well as with the canon 200mm lens. The iris nebula on the other hand I wouldn't consider as easy a subject as the rest, given my FOV and the fact that most of the interesting artefacts around the nebula consist of VERY FAINT dust lanes. In any case, I decided I wanted to take the leap and just go for it. 

A concern of mine was the colour of the stars; I wanted to make sure that I didn't clip any of them, and so I decided to expose my subs to accommodate this, which lead me to 240s subframes at ISO1600, dithered. This is what the histogram looked like for one sub:

And this is what a single sub looked like without any processing/calibration:

 

The stars seemed sharp enough and I had both the Iris in view as well as the ghost nebula (just about) keeping it company. My train of thought has always been that even if a single sub doesn't show the detail I'm after, stack enough of them and it'll soon appear. I went ahead and captured the images, got the corresponding flats, darks and bias frames and then ultimately started to process the images. 

This is what the histogram looks like in PI before and after I apply the DEFAULT screentransferfunction curve:

And this is what the image looks like, without any post-processing other than Automatic Background Extraction and a 180 deg fast rotation:

There are a number of things that concern me about this image.

1. The colour motle/noise what ever you want to call it. There are blotches everywhere.
2. The level of luminance noise which for 8+hours worth is simply disappointing.
3. The lack of detail in the dust lanes.

I come back to the title of this thread:

Where did I go wrong? 

I apologise for the length of this post, I was just trying to be thorough.

[johnrt]

You have done nothing wrong, those molecular dust patches are very difficult to do well, I struggle to capture them with my mono CCD, so with a DSLR you are going to need some dark skies to get them.

The background cleans up nicely with a bit of know how & I found your black point to be way too high. A little bit of processing starts to show some really nice detail, here's what I could do in 60 seconds from your jpeg.

 

[Shibby]

Yes, the main 'problem' will be the signal-to-noise ratio of a 4 minute ISO1600 exposure. An 8-minute ISO800 exposure will be far better for capturing the incredibly faint dust around the Iris. (Of course, it's a bit more complicated than that. I'm not sure how 50 subs of one should compare to 25 of the other, for example.) Actually, low ISOs are also better for maintaining star colour.

[Rico]

38 minutes ago, johnrt said:

You have done nothing wrong, those molecular dust patches are very difficult to do well, I struggle to capture them with my mono CCD, so with a DSLR you are going to need some dark skies to get them.

The background cleans up nicely with a bit of know how & I found your black point to be way too high. A little bit of processing starts to show some really nice detail, here's what I could do in 60 seconds from your jpeg.

 

Johnrt, the image definitely looks a lot better with those additional touches and you're right in that processing will make a huge difference. I'll see if I can spend some extra time on this maybe I'm being too difficult on myself?

[Rico]

24 minutes ago, Shibby said:

Yes, the main 'problem' will be the signal-to-noise ratio of a 4 minute ISO1600 exposure. An 8-minute ISO800 exposure will be far better for capturing the incredibly faint dust around the Iris. (Of course, it's a bit more complicated than that. I'm not sure how 50 subs of one should compare to 25 of the other, for example.) Actually, low ISOs are also better for maintaining star colour.

Ironically, before capturing I was toying with the idea of doing 8 minute ISO800 exposures, which would have given me the same histogram with a better SNR, but I was a little worried about the clouds, so I figured I would keep to ISO1600, expecting not to bin as many subs. SNR is definitely something I need to think more about, and my impression at least is that while the iris is well above the readnoise, the dust lanes are not. Maybe I should have exposed for the dust lanes by doing some long exposures and then exposing for the stars, combine the two and have a far more balanced image. I think my mistake was treating the image as one object when in fact it is composition of many.  

[Rico]

@johnrt how did you manage to get rid of the huge rgb blotches scattered in the image. I've zoomed into the image to give you a better idea of what I mean:

Thanks

[AKB]

1 hour ago, Rico said:

how did you manage to get rid of the huge rgb blotches

If you take the green away (with SCNR or similar) then they look a lot less offensive...

[ollypenrice]

You have probably chosen this excellent target because you've liked images you've seen on the net? That's fine, but did you check out the kit used to take them? And the integration time? I think you've done pretty will with the kit and the time. For me this is a 20+ hour project from a dark site with cooled CCD. 

This morning I saw the same target in a 'quick and dirty' process from one of the robotic rigs I host and a new CMOS cooled OSC camera. It may be a game changer...

Olly

[Rico]

30 minutes ago, ollypenrice said:

You have probably chosen this excellent target because you've liked images you've seen on the net?

Absolutely! If it weren't for those amazing captures, including the ones that you yourself have shared with the community, I wouldn't know what to expect when I look at the sky, let alone where to look. I think more than not I shy away from objects like these, because of the equipment that I have and the light pollution from nearby cities. But before I jump on the CCD/cooled CMOS wagon and invest more and more in this hobby, it makes sense to me so find the limitations of what I have and I guess what I can do. This seemed like a perfect candidate. 

What I really needed to figure out is whether I've made a mistake in the manner in which I've calibrated/processed the image or acquired it. The huge red/green blotches scattered in the image, which to some extent are tackled by @AKB 's suggestion of using a SCNR is what bothers me the most. @johnrt seems to have dealt with them quite nicely in his version albeit with a jpeg version.

 

[AKB]

2 hours ago, ollypenrice said:

a new CMOS cooled OSC camera. It may be a game changer...

Good grief!  I never thought I see you write that!  You'll be extolling the virtues of Crayfords next...

[ollypenrice]

9 hours ago, AKB said:

Good grief!  I never thought I see you write that!  You'll be extolling the virtues of Crayfords next...

No, I've always been open minded on CMOS and, so far, this is only available in OSC. 36 meg and full frame has to tempt any FSQ owner. Mono would be perfect!

Olly

[gorann]

3 hours ago, ollypenrice said:

No, I've always been open minded on CMOS and, so far, this is only available in OSC. 36 meg and full frame has to tempt any FSQ owner. Mono would be perfect!

Olly

Olly, does that OSC have a name?

[ollypenrice]

29 minutes ago, gorann said:

Olly, does that OSC have a name?

I'll PM you to avoid thread derailment.

Olly

[wimvb]

On 2017-10-18 at 16:00, Rico said:

The colour motle/noise what ever you want to call it. There are blotches everywhere.

Chrominance noise reduction. Either mmt on 8+ layers in the linear stage, or acdnr at the end of the workflow (nonlinear). Both with a proper luminance mask. You ca also try a masked or arcsinh stretch on rgb, and a separate, more aggressive stretch on lum (extracted from the linear rgb image). Then use lrgb with chrominance noise reduction activated.

And as said previously: a dark site is almost essential for this target.

[Rico]

Thanks for the suggestions. I had another go at processing the image. It's definitely made me realise how much I need to develop my own processing skills. I'll hopefully add another night's worth of data and call it a day. In the meantime ... I need to make a CCD/CMOS shopping list. But that's for another thread ... .

[cfpendock]

That is turning into a really good image.

Chris

 

[ollypenrice]

You originally asked what you were doing wrong and John, who knows what he's talking about, (see his images!) said, 'nothing.' He was right!

Processing is a never-ending learning process. The opportunity to learn will literally never end.

Olly

[Rico]

Thanks for the vote of confidence (@cfpendock @ollypencrice). Now, I play the waiting game ... with storm Brian now approaching it'll be a while before I'm outside imaging. 

[StuartJPP]

1 hour ago, Rico said:

Thanks for the vote of confidence (@cfpendock @ollypencrice). Now, I play the waiting game ... with storm Brian now approaching it'll be a while before I'm outside imaging. 

Don't worry, I have heard that the Iris nebula will still be there next year

Good image and getting the dust is tricky so you have done well....

 

[MartinB]

As far as I can see your data is excellent.  Optimum sub exposure times are determined by your read noise and your skyglow.  A low read noise sensor from a light polluted site won't benefit from over long subs.  Once background skyglow is above your read noise threshold stacking will take care of your signal to noise.  My experience is with CCDs and there are good calculators around to determine optimum sub length, I'm sure there are similar for cmos chips.

PI makes even the simplest processing actions sound incomprehensible!  The automated screen stretch function applies a vicious curve and John's more gentle touch and your subsequent rework looks to be delivering a great result.