Heart Nebula Help

[Rodd]

Greetings all.  I am bumping into that omnipresent wall...frustration!  For a little history,  I am new to imaging (been at it about 1 month), and am finding the processing learning curve to be somewhat overwhelming--beginning with what software to use.  I use Nebulosity and PixInsight for the most part (actually I have the PixInsight 45 day free trial  and am trying to decide whether I should buy it--it expires in 2 days!).  I also have AIP$win 2.0.  I have trouble with the concept of a monthly subscription (PS) as over 10-20 years it will add up to a huge amount.  However, I'm beginning to think it may be the way to go.   The following pics are of the Heart Nebula taken with a TeleVue np101is (.8x reducer), Astrophysics Mach 1, SBIG STT 8300.  The image consists of HaRGB.  Note Ha was used as the L channel only.   Total Exposure time was 14.7 hours (Ha: 11 x 20 min, R: 11 x 20 min, G: 11 x 20 min, B, 11 x 20 min) at a temp of -45C.  I used a master dark composed of 21 20 minute dark frames at -45C.  The image was calibrated, aligned, normalized and stacked in Nebulosity.  I did stretch the image in Nebulosity (saturation, color, contrast etc), but that's about it. Then I Used TGVdenoise and deconvolution in PixInsight.  Any recommendations would be appreciated.  Many thanks (The 1MB limit for picture upload has me a bit baffled as it forces me to use a highly compressed JPEG file.  AP images are more suited for TIFF or FITs and data can be lost (at least that's what PixInsight says when I save as a JPEG).  One point--guiding was very good (less than .2 pixel errors on x and y graphs.  The lines visible in the pic are artifacts of the nebulosity (either good ones (real) or bad (artificial), I don't know) created/revealed by the TGVdenoise/deconvolution process. 

[StargeezerTim]

You have produced this after only a month!

Thats superb, no phenomenal, progress...

[laudropb]

Lovely image. You have certainly got the hang of the dark side quickly.

[Barry-Wilson]

Hi Rodd

Processing is a skill to acquire and it is good that you have recognised this.  Calibration is the basis of processing and you mention dark frames but not flats or bias - have you produced these too?  It is essential to have well calibrated files to maximise the potential of your data.

I use PixInsight and there are good web-based tutorials to help you get started: Light Vortex Astronomy, http://www.lightvortexastronomy.com/p/tutorials.html;and Harry's Astro Shed, http://www.harrysastroshed.com/pixinsight/pixinsight%20video%20html/Pixinsighthome.html.  I'm sure there are good tutorials for Photoshop but I'm not familiar with those however others will post up references I'm sure.

Irrespective of which software you use there is a common basic workflow for the principal steps when processing an LRGB or HaRGB image.  Your Ha files alone will produce a satisfying image I'm sure - Ha monos of this target are a lasting favourite.

In PixInsight, it starts with Batch Pre Processing, which calibrates and aligns your lights.  You will need darks, biases and flats as well as your lights.  I would also recommend using Cosmetic Correction too.  Harry's tutorials explain this process well.

Harry's tutorials also explain the following steps, along with basic tools like STF and Histogram Transformation, saving files etc:

1. As a basic process you could use the Master RGB and Ha files (output reslts from the Batch Pre Processing) rather than align the calibrated individual FITs (but that is preferable and you use Image Integration on the calibrated files to do this).

2. Thereafter you will need to combine the RGB using Channel Combination.

3. Crop the RGB and Ha files using Dynamic Crop (saving the instance to apply to the uncropped files)

4. Then carry out a Dynamic Background Extraction on the RGB and Ha.

5. Do not use Deconvolution - it is a more complex process that you can use when your experience grows.  But for now the risks of producing a poor result probably outweigh the benefits.

6. The Ha can be stretched simply by transferring the STF instance to the Histogram.

7. You will need to carry out Background Neutralisation and Colour Calibration on the RGB file.

8. Once you have colour balanced the RGB I would carry out a masked stretch with default settings (you may experiment but this is your first image with PI).

9. Then combine the RGB and Ha stretched files using the script HaRVB - AIP in the Script/Multichannel Synthesis menu.

10. Use Histogram Transformation or Curves to refine your image, including saturation boost.  Small steps that you may repeat are better than one large change.

11. You may need to use SCNR with a green setting to remove any trace of a greenish wash in the image.

This is a very basic workflow that will get you an image.  The above two tutorial web pages will give you a lot more information and other techniques which you may want to use.

I'm sure you can improve on your image above as you have degraded a lot of detail which I am sure is present in your data.  Your colour balance is not correct either.

There is a lot of information in the SGL forum pages 'Imaging - Image Processing, Help and Techniques' and I'm sure folk will gladly volunteer advice.

Good luck and enjoy!

[Rodd]

Hi Rodd

Processing is a skill to acquire and it is good that you have recognised this.  Calibration is the basis of processing and you mention dark frames but not flats or bias - have you produced these too?  It is essential to have well calibrated files to maximise the potential of your data.

I use PixInsight and there are good web-based tutorials to help you get started: Light Vortex Astronomy, http://www.lightvortexastronomy.com/p/tutorials.html;and Harry's Astro Shed, http://www.harrysastroshed.com/pixinsight/pixinsight%20video%20html/Pixinsighthome.html.  I'm sure there are good tutorials for Photoshop but I'm not familiar with those however others will post up references I'm sure.

Irrespective of which software you use there is a common basic workflow for the principal steps when processing an LRGB or HaRGB image.  Your Ha files alone will produce a satisfying image I'm sure - Ha monos of this target are a lasting favourite.

In PixInsight, it starts with Batch Pre Processing, which calibrates and aligns your lights.  You will need darks, biases and flats as well as your lights.  I would also recommend using Cosmetic Correction too.  Harry's tutorials explain this process well.

Harry's tutorials also explain the following steps, along with basic tools like STF and Histogram Transformation, saving files etc:

1. As a basic process you could use the Master RGB and Ha files (output reslts from the Batch Pre Processing) rather than align the calibrated individual FITs (but that is preferable and you use Image Integration on the calibrated files to do this).

2. Thereafter you will need to combine the RGB using Channel Combination.

3. Crop the RGB and Ha files using Dynamic Crop (saving the instance to apply to the uncropped files)

4. Then carry out a Dynamic Background Extraction on the RGB and Ha.

5. Do not use Deconvolution - it is a more complex process that you can use when your experience grows.  But for now the risks of producing a poor result probably outweigh the benefits.

6. The Ha can be stretched simply by transferring the STF instance to the Histogram.

7. You will need to carry out Background Neutralisation and Colour Calibration on the RGB file.

8. Once you have colour balanced the RGB I would carry out a masked stretch with default settings (you may experiment but this is your first image with PI).

9. Then combine the RGB and Ha stretched files using the script HaRVB - AIP in the Script/Multichannel Synthesis menu.

10. Use Histogram Transformation or Curves to refine your image, including saturation boost.  Small steps that you may repeat are better than one large change.

11. You may need to use SCNR with a green setting to remove any trace of a greenish wash in the image.

This is a very basic workflow that will get you an image.  The above two tutorial web pages will give you a lot more information and other techniques which you may want to use.

I'm sure you can improve on your image above as you have degraded a lot of detail which I am sure is present in your data.  Your colour balance is not correct either.

There is a lot of information in the SGL forum pages 'Imaging - Image Processing, Help and Techniques' and I'm sure folk will gladly volunteer advice.

Good luck and enjoy!

Thanks everyone--I appreciate the comments.  Wow Barry--I'll have to dig into this list.  Understanding it and doing it are 2 different things.    I have not made flats or biases yet because without flats bias frames are useless (so I've been told), and flats are a bit of a mystery to me--I purchased a flat image thing that goes over the scope.  It hasn't arrived yet.  Making a flat box is beyond me.  Hopefully the thing works.  I do have Warren Keller's 3 video tutorials for PixInsight--but the batch processing is not available on the 45 day free trial.   I have found Keller's videos informative, but difficult to use as they are so fast and sometimes it is difficult to know when to ignore stuff (like debayering color images since I use a mono camera with filters).  I sort of jumped into deconvolution because quite frankly, it was the first thing I tried that actually produced results that didn't completely destroy the image.   

I am mystified about the image quality--especially when enlarged.  When small, the image has caused people to say wow--decent image for a start.  But when you enlarge it, it drops into the pathetic range-not much more than a cartoon.  This came to light when I tried to print it out to give to my wife as a Christmas present (a nebula that really does look like a heart is a romantic thing--softening the blow of spending many hours alone whenever its clear out!).  The printer said that the image is 72 dpi--the camera I use is 8.3 megapixels and should produce higher resolution images.    I have seen stunning high resolution images taken with the same equipment.  It would be nice to get at least a glimpse of it in an image that has over 14 hours of exposure.

Many thanks,

Rodd

[Barry-Wilson]

Great, you have the Warren Keller dvds - I have these too and they are a good resource.  Harry's tutorials are also a terrific place to start as a beginner.

If you have purchased the dvds, it seems sensible to acquire the full version of PI.

Flats are invaluable and you can make sky flats using a t-shirt to cover the objective or point your objective at your computer screen with desktop colour of a dull grey.  The book 'Making Every Photon Count' by Steve Richards is indispensible, http://www.firstlightoptics.com/books/making-every-photon-count-steve-richards.html.

HTH

Barry

[swag72]

I agree with Barry that flats really are a must and as for who ever told you bias were useless I think has the wrong idea. 

I think that you've done something that we've all been very guilty of and that is over processing. I would tackle this in the following way.

• Calibrate and stack your individual images into stacks for each filters.
• Combine them into the appropriate channels
• Do a standard curve (This is shown on my website - Unfortunately I am not allowed to put my website links on SGL - It's under More --> My guide to image capture --> Sart / finish processing is the key ..... This shows the first three standard astro curves in Photoshop and will get you seeing something.
• Adjust the levels and do not clip any of the data
• Continue to slowly stretch the data as appropriate

What I have learnt is that it's all touchy feely and what works for one image will not work for another. From there I'd be looking at the image and what it may need - Some saturation boost, noise reduction etc.

[Rodd]

Thanks Sara.  There is no doubt I have over processed.  However, the unprocessed image is grainy and is even lower resolution.  Here it is for comparison.

[swag72]

Are you saying that this image you have posted above is unprocessed? By unprocessed I mean that it has been combined into the correct RGB channels and then nothing else? It should appear almost black - That's called a linear image

[Rodd]

No Sara,-when I say unprocessed I mean no "advanced" processing.  I did power stretch and play with contrast and saturation, curves--the simple stuff that is easy to do.  The stuff I can do to make the pic better.  Beyond this, anything I try ruins it-stuff I don't even understand with multiple settings and sub decisions.  I have seen some of your pics--the crops--reminiscent of the Hubble telescope images.  They look real.  Amazing.  Obviously way beyond a mere flat subtraction.  

Rodd

[swag72]

I think honestly Rodd you should take it back to basics - I have no idea what a power stretch is, but I'd be stretching in small iterations not all in one go. Forget advanced processing until you have the basics right - This is not quite right as it stands

Hope that helps

[Rodd]

That's one of the problems I face--many different processing software platforms--many different terminologies, different users using different platforms.  Me not knowing which is best, or if they are equal which to buy.  Power Stretch is a Nebulosity term--probably just means stretch.  You know, a 3 prong approach to stretching luminance, contrast, and something else I don't really know what it is.  Actually in Nebulosity there is a 3 prong approach to adjusting background color, color balance, luminosity and contrast.  There is alos a automatic color balance (which I used).   Its all just manipulating the levels in a subjective manner until one is satisfied.  I Guess PixInsight is more objective, but the complexity is severe.  But, my point is I have spent literally hours manipulating those curves/stretches--what ever you want to call them, and the answer does not lie there.  One might not like the color balance, or contrast or whatever, but no version of my "stretching" was better than this one.  There is something more, something deeper that needs to be done.  Noise reduction, masks (unsharp or stars etc.), on and on and on.  At present I am at a point where I have reached my ability to positively impact an image using the techniques that I have become familiar with.  There is a tool (probably many) in the tool box that I don't have.  I have manipulated as many elements as I can in efforts to make the picture as clear as possible.  As high resolution as possible.  I have many dozens of end results of this pic.  All started with calibration, normilization, algnment and stacking.  Granted Flats and Bias frames were not used.  They would help I am sure--but I don't think they would be that noticeable in this pic.  They certainly would not transform it into a high resolution image.

Thanks,

Rodd

[ChrisLX200]

It's difficult looking at your final result how you got there, and just trying to guess what the raw data looks like as a starting point is not helping us to come up with answers. For sure the colour is just all wrong, there are colour biases resulting in a strong red in the top-left, and a strong green in the bottom-right, plus the highlights are blown so the details in those areas are gone. The bottom-left suggests the levels are clipped to black too. So, as Sara suggests above, you need to go back to basics and process the channels gently without excessive stretches with the aim of maintaining a balance between the RGB channels. You are using the H-alpha as luminance so the detail and sharpness will all be in that channel.  It would help to understand what the particular problems are if you can post the unprocessed stacks.

ChrisH

[Rodd]

Here's the Ha stack concerted to JPEG.  I'll have to post each stack on different postings due to size constraints (that's a pain).  Blue will be next, then red then green

[swag72]

This looks processed to me .... Is your software doing some sort of auto stretch? 

[Rodd]

Blue:

[Rodd]

Green

[Rodd]

Red

Wow--Many thanks are in order.  Truly.

[Rodd]

Sara--I don't know--I don't think so.  When the subs pop up when the exposure time is done at the scope the Ha subs looked like this pretty much

[johnrt]

Is this a mix of Ha and RGB? If so how have you applied the different captures together to form the image?

What do you mean when you say "high resolution" image?

Your Ha mono image looks decent, I expect something is going wrong with how you combine the colours together.

Your capture software is most likely auto stretching for you, a totally unprocessed image will be very black.

[swag72]

Your capture software is most likely auto stretching for you, a totally unprocessed image will be very black.

John is right - These are processed to some extent - An unprocessed linear image WILL be almost black. You need to make a start with UNPROCESSED images.

[Rodd]

I us ethe Ha for the Lum channel, the Red stack for the Red channel, the Green Stack for the green channel and so on.  I keep a 1:1:1 color balance so as not to change anything.  

[ChrisLX200]

Here's the Ha stack concerted to JPEG.  I'll have to post each stack on different postings due to size constraints (that's a pain).  Blue will be next, then red then green

OK, that stack is already stretched? I'm pretty sure the ST8300 has enough well-depth that the highlights would not be blown as they are shown here. As said, this channel will contribute all the detail and sharpness (and noise if there is any!) so it has to be your best in that regard and the RGB will basically just 'colourize' what this channel is showing. You are free to process this channel independantly - the RGB on the other hand need to balanced with each other.

ChrisH

[Rodd]

I did not do anything except calibrate (without flats and biases), normalize, align and stack in Nebulosity.  I know, Nebulosity is probably not in your bag of tricks--it is very popular in the States--that is why I got it.  

[Davey-T]

Brighter parts of the HA have been stretched beyond redemption unless it's just overexposed are you sure it hasn't been stretched ?

Dave