Help needed

[CotswoldsBloke]

Hi

 

After much anticipation, I finally got out last night for a clear night. My aim was to photograph the Heart Nebula using my ZWO ASI533 MC PRO, which was attached to my Redcat 51 & Optolong L-Extreme filterEverything was linked to a ZWO ASI PRO and everything worked as it should in terms of finding the object, plate solving. I had the camera set to fire 30 x 5 min exposures with a gain of 100. When the first exposure came through, I could clearly see the red of the nebula and nice pinpoint stars. I left the rig running and was very excited to see 29 frames all showing a nicely centred image. All good.

The issue comes when I come to stack the images in Deep Sky Stacker. When I check the light frames, I cannot see the nebula itself even with aggressive stretching. In fact, it really doesn't show very much at all, even the stars are very muted. I have stacked the image, but the histogram is showing very little data.

I am a relative newbie to this and I'm struggling to understand why the image looked decent on the ZWO preview/review but seems almost devoid of data in DSS and Photoshop. Anyone have any ideas?

Thanks

Nick

[vlaiv]

Could you attach 3 things for inspection:

- first raw frame of the night (without calibration)

- last raw frame of the night (without calibration)

- linear stack from DSS as 32bit fits format

There might have been dew forming on your scope. That will ruin stacked image. If that is so - there will be significant difference between first and last frame (clear lens and dewed up lens).

[CotswoldsBloke]

Thanks for replying.

I've attached as requested.

Light_IC1805_300.0s_Bin1_0001.fit Light_IC1805_300.0s_Bin1_0030.fit Autosave.tif

[vlaiv]

3 minutes ago, CotswoldsBloke said:

I've attached as requested.

Good news is that it's not dew. Both first and last sub of the sequence show nebulosity clearly:

Last sub has more LP - but that is probably due to direction of the target in the sky as it moves thru the night.

I'll now inspect stack to see what it looks like.

[vlaiv]

Ok, stack also shows that data is there and all is "fine":

Problem is of course with light pollution and fact that stack has very high dynamic range - it needs to be highly stretched to show faint stuff.

You have quite decent data - now you need to put effort into processing and learn how to best pull the image out of it.

Above is just a quick manipulation in Gimp, if you wish, I can do a bit more serious processing of this data to see what I can pull out of it?

[vlaiv]

Hope you don't mind - I took the liberty to process your data more seriously and here is the result:

[CotswoldsBloke]

vlaiv, do I mind? Absolutely not! I am amazed that you've gone to such lengths to help me. Seriously, I can't thank you enough.

I am just pleased that I managed to get good data, which is a step forward for me. The processing side is still mysterious to me, but seeing what you've done with my data only encourages me.

I just need to work out how to do it myself as my initial efforts of levels and curves produced an absolute mess.

Edited May 27, 2021 by CotswoldsBloke

[alacant]

2 hours ago, CotswoldsBloke said:

DSS and Photoshop

Hi

Try other software perhaps. You have some wonderful detail captured there. 

Cheers

[CotswoldsBloke]

Wow. 

[vlaiv]

2 hours ago, CotswoldsBloke said:

I just need to work out how to do it myself as my initial efforts of levels and curves produced an absolute mess.

I have no problem showing you some of the steps I took while processing this data.

One part is a bit more involved - that is background removal. I use ImageJ and custom written plugin to do that. There are tools that you can use that will wipe background for you (like gradient Xterminator for PS that I've heard of but have not used it).

In the mean time - to practice, I'll upload data that I wiped so you can use it.

I binned your data x2 - as you are using OSC sensor and shooting with very restrictive filter. Only red and a bit of teal (OIII / 500nm) is passing thru - and there is not much data in green and blue channel:

(this is R, G and B data from your image side by side - blue is almost non existent).

Bayer matrix on sensor has every fourth pixel as red (RGGB 2x2 group of pixels) and there is not much point in having 3000x3000 image in mostly red color - you gain no detail by using such resolution.

Data is also wiped / normalized:

red.fitsblue.fitsgreen.fits

You can now perform simple processing using Gimp for example (or PS - just use the same technique). Here is very simple 3 step level stretch, a bit of curves manipulation and some noise reduction:

- Load channels and do RGB combine to get color image

- Start by doing one round of levels - moving only top slider - move slider all the way down until you start clearly seeing nebulosity. This method will make all star cores bright and is not suitable if you want to retain star color - but it is very simple technique to use and it works good on image like this where star color is mostly lost due to filter used.

so move top slider to the left - until you start seeing bright features of nebulosity - don't over expose / burn bright part of nebulosity though. Click ok.

2. Now we move middle slider in next step - we move it to the left again - this brings forward fainter parts of image but also raises background level (don't worry about that)

Again click ok when happy.

3. You need to adjust black level on the image - raise left most slider up until the base of histogram. You can also set slightly higher output level - so that background does not look completely pitch black - that does not look good in images:

4. Put a bit more emphasis on nebulosity using curves now instead of levels like this:

put one anchor at center of histogram so that this part does not move and use another adjustment point to the right so that you get nice sloping curve - adjust to your liking.

5. Do some noise reduction now. I use simple wavelets noise reduction - here is comparison - left side is unaltered and right has been denoised:

Difference is obvious. You can also do masked denoise - only in dark parts (copy layer - denoise that, add layer mask with luminance as mask - bright parts should show original bottom layer, dark parts of mask should show denoised layer).

Hope this gets you started in processing. You can also check out some videos on YouTube.

[CotswoldsBloke]

That's very generous of you. I will certainly run through it and practise.

I had a go myself following a tutorial using Pixinsight. It's way off what you and alacant achieved, but it's a start!

What a great day - I've learned a whole lot and got a recognisable image too.

Heart Nebula 270521.tif

Heart Nebula 270521.tif

Edited May 27, 2021 by CotswoldsBloke

[vlaiv]

3 minutes ago, CotswoldsBloke said:

I had a go myself following a tutorial using Pixinsight. It's way off what you and alacant achieved, but it's a start!

That is excellent - and in no way lacking in comparison to our processing efforts.

[scotty38]

Just wanted to say what fantastic support provided on this thread, well done!

[alacant]

5 hours ago, CotswoldsBloke said:

initial efforts of levels and curves

Hi

Your effort is fine and fortunately these days we have a choice. Modern software gives us the possibility of moving away from levels and curves altogether.

Cheers

 

[CotswoldsBloke]

2 hours ago, alacant said:

Hi

Your effort is fine and fortunately these days we have a choice. Modern software gives us the possibility of moving away from levels and curves altogether.

Cheers

 

Which tools do you prefer? It would be good to understand why, too. 
 

Thanks

[vlaiv]

I would like to know "why" part as well

 

[alacant]

1 hour ago, CotswoldsBloke said:

Which tools

Hi

We prefer StarTools. A breath of fresh air after the ubiquitous levels and curves approach which hasn't changed since last century. It keeps your data linear as you process. You are not restricted by having to do things in the age-old correct order. Or bust. You work with a clean screen devoid of a mess of windows. It screams on most hardware with its groundbreaking GPU processing support. You can get a decent image quickly and simply without having to type numbers onto a graph or 'get the background right', whatever that is. You don't need to spend hours redoing stuff. 100 other reasons omitted here...

Perhaps best if it's author @jager945 describes his philosophy and why it works so well. Or just have a read: https://www.startools.org

HTH

Edited May 27, 2021 by alacant

[The Lazy Astronomer]

26 minutes ago, alacant said:

Hi

We prefer StarTools. A breath of fresh air after the ubiquitous levels and curves approach which hasn't changed since last century. It keeps your data linear as you process. You are not restricted by having to do things in the age-old correct order. Or bust. You work with a clean screen devoid of a mess of windows. It screams on most hardware with its groundbreaking GPU processing support. You can get a decent image quickly and simply without having to type numbers onto a graph or 'get the background right', whatever that is. You don't need to spend hours redoing stuff. 100 other reasons omitted here...

Perhaps best if it's author @jager945 describes his philosophy and why it works so well. Or just have a read: https://www.startools.org

HTH

Plus it's crazy cheap compared to other popular paid-for options (APP, Pixinsight, Photoshop).

I found it does take a bit learning to get to know the right settings for your data (I find I have to tweak the parameters around a bit to get a good final stretch I'm happy with - however this mostly due to my poor quality data) and fortunately Ivo is more than happy to help if you contact him 👍

[vlaiv]

2 minutes ago, The Lazy Astronomer said:

Plus it's crazy cheap compared to other popular paid-for options (APP, Pixinsight, Photoshop).

Not sure if it is fair comparison.

APP and PI are both capable of stacking the data and PS is full fledged image processing application.

One could alternatively argue that Gimp is free - so even cheaper.

[The Lazy Astronomer]

35 minutes ago, vlaiv said:

Not sure if it is fair comparison.

APP and PI are both capable of stacking the data and PS is full fledged image processing application.

One could alternatively argue that Gimp is free - so even cheaper.

Yeah, fair point - I actually forgot APP and PI were stackers as well. And also that 'normal' people might want to do something that isn't just astro image editing 🙃

[alacant]

7 hours ago, The Lazy Astronomer said:

APP and PI were stackers as well

No problem.

Stacking? Siril is free,  just as powerful and also has a modern user interface.

Cheers

Edited May 28, 2021 by alacant

[Clarkey]

I think in most cases you can trial the various bits of software which is worth doing and comparing your results.

Personally I use a number of different options including Startools and more 'traditional' software. Both have advantages and disadvantages. One benefit of the more traditional approach is to understand what and why you are doing certain things and the effect they have. As Alacant says, Startools does most of the same things in a different way - but I found Startools easier once I had used APP and Affinity as I new what I was trying to achieve at each stage of processing.

[bottletopburly]

If you struggle with the likes of Gimp , photoshop,affinity I certainly did ,levels and curves taking forever and a day to get something then I would highly recommend Startools , +1 for Ivo being super helpful should you need guidance.

[The Lazy Astronomer]

4 hours ago, alacant said:

No problem.

Stacking? Siril is free,  just as powerful and also has a modern user interface.

Cheers

I've never tried Siril, how does it compare to DSS?

[alacant]

43 minutes ago, The Lazy Astronomer said:

never tried Siril, how does it compare to DSS?

Hi

IMO, it's better by design. You tell it what you want, rather than vica versa. It has an excellent choice of alternative stacking algorithms and comprehensive debayer routines along with some very good background and noise reduction modules.

Recommended:)