NGC 7000 (The North American Nebula) with IC5070 (Pelican Nebula) in Narrowband- Total of 9.33 Hours

Date: August 11, 2021

Cosgrove’s Cosmos Catalog ➤#0081

NGC 7000 - The North American Nebula with IC 5070 - the Pelican Nebula towards the bottom-right. (click image for full resolution via Astrobin.com)

Published as a two-page spread in the October 2022 Issue of Sky & Telescope!

Published in Amateur Astrophotography Magazine!

April 9th, 2023 - In issue #111, p70-75. This image was published as part of a profile article on me!

Table of Contents Show (Click on lines to navigate)

About the Target

NGC 7000, also known as the North American Nebula and Caldwell 20, is a large H II emission nebula located in the constellation of Cygnus (The Swan). There has been some controversy around its distance from Earth. Still, recent observations of the Gaia Astrometry Satellite of 396 stars within the nebula have nailed this down to 2590 light-years away. It measures roughly 2 degrees by 1.5 degrees in size, making the nebula about 10X larger than the area covered by the full moon.

It was first discovered visually in 1796 by Sir William Hershall and was added to the New General Catalog as NGC 7000 by his son, John Hershal, in 1829. Early long-exposure photographic plates showed NGC 7000's characteristic shape, the genesis of the common name we know it by today.

Open Cluster NGC 6997 can also be seen towards the middle of the field. While this cluster looks embedded and part of the nebula, it is actually not associated with it at all - instead is located hundreds of light years beyond the nebula itself.

Also shown in the field is the Pelican Nebula, IC 5070, seen on the bottom-right side of the image.

The Annotated Image

The annotated image was created using the ImageSolver and AnnotateImage scripts in Pixinsight.

The Location in the Sky

This IAU/Sky & Telescope Constellation Chart shows the location of the North American Nebula indicated by the yellow arrow.

About the Project

I have always wanted to shoot this target, but it is so large that my first two telescope platforms could only cover a tiny portion of it. So for a long time, I really had no widefield capability.

But my third platform, the Askar FRA400, is perfect for this! I captured this over three nights, starting August 2, 2021, when we had a rare (for this summer) clear patch with no moon, clouds, or smoke from Western fires.

The framing was done with the Framing and Mosaic Wizard in Sequence Generator Pro (SGP). Since the Pelican Nebula was close by, I wanted to include it in the field of view. With the right rotation, I was able to get them both in the frame.

But - I have no camera field rotator on this Platform. Normally, SGP would do a centering operation before starting a sequence. First, it would do a platesolve and then move the scope to point at the right coordinates. Then it would determine the amount of camera rotation needed to get the framing right - and finally would move the rotator and confirm positioning.

So how does this work when there is no Rotator? Well - I become the Rotator! I connect SGP to a "Manual Rotator." When it does its centering operation, it will pop up a message that says, to the effect: "Hey dummy, rotate the camera 23.4 degrees", and then it does another platesolve. "Nope - I need you to rotate it another 4.3 degrees." When it is happy, it moves on to the next step.

Exposures over three nights seemed to go smoothly, and the tracking of the Ioptron CEM 26 has been excellent.

Image Processing

In the past, I have not really covered what I did for my image processing. With this new image, I am changing that. I took notes while working on this image, and I will share the exact steps I took to process this Image.

• Those of you that are not involved in processing can skip this part.

• Those of you that are trying to learn dig in!

• Those of you who are experts at this and wish to critique my processing decisions and tell me that I am doing it all wrong should feel free to send me an email at take-a-flying-leap@cause-I-dont-care.com.

• Those that have constructive suggestions - I am always eager to learn more. Use the address on the contact page, not the one listed just above. ;-)

Under-Sampling

In the most recent images I have done with the portable Askar FPA400 platform, I have seen that my images are under-sampled. What does this mean? It means I really need a higher resolution camera for the image scale my system has. Normally a star's light falls on several pixels. If your sensor does not have sufficient pixel density, it may fall on, say… one, or two pixels. So when you are under-sampling, your stars might be squarish or perhaps a single pixel. Not good.

So there is a method of fixing this - it involves using a credit card and buying a new camera. While this is a simple method that will clearly fix one problem, it may create others (mad wife, divorce, etc.), so I will stay away from that solution for now.

There is a method called Drizzling. This is a method first developed and used by the folks at the Hubble Space Telescope. It’s a way of taking multiple frames and wringing for information from them. A simplistic explanation is that the image data is rotated and placed over a virtual array of pixels with a higher resolution. Then, the new pixels are computed based on that overlap.

Here is a chart from the Drizzlepac Handbook 2.0 from the Space Telescope Institute illustrating the concept.

You can read this handbook yourself and learn how the Hubble folks use this technique HERE.

Pixinsight supports Drizzle processing, and this was my first time using this method. It seemed to work well on the plus side, and the image created with it looked good. But, unfortunately, the newly created image has a whopping resolution of 9126 x 6878! That’s a lot of pixels, and processing took a hit because of that (even my 12-core image processing beast struggled with it!)

So below is my processing log with notes.

---

NGC 7000 Processing Log

1. Blinking the Images

• All lights and cal frames were blinked

• I only found one issue, a single Ha sub was removed because it seemed to have much higher exposure than all others - not sure what happened here, but the stars were super blown out.

• Data looked super clean and very consistent frame to frame. See the video below of the Ha frames from 3 nights shown at the rate of one-frame-per-second. (Note that you will see orientation flip 3 times - one Meridian-Flip per night!)

2. WBPP v2.1 was loaded with all subs and cal frames

• The Drizzle option was chosen

• A value of 50 was used as a pedestal image as this was narrowband

3. DrizzleIntegration was used to create Ha, O3, and S2 master images.

4. DynamicCrop - all master images were cropped in a consistent fashion to exclude ragged edges.

5. DBE was NOT run.

• The whole image was covered in the nebula, and no obvious gradients were visible, so I skipped this step.

6. Prep for Deconvolution

• Object Masks created for each layer. HT was used to adjust so that the background sky was black, and most Nebula areas were white.

• Local support images were created. These are basically star maps of the biggest and brightest stars. Since these are usually saturated and clipped, they never have normal point spread functions, so they don’t fit the deconvolution model and it does bad things to them. I used Starmap with 6 layers and every else set to default for this.

• Psf files created- These are point spread function files. I created the using the PSImage Script. Piece of cake!

7. Run Deconvolution

• Setup several preview areas to test on each image

• Apply the object mask

• Set the deconvolution tool to use the right psf and local support maps

• For each layer - explore what global dark value gives the best response without dark rings.

  • Ha - global dark of 0.02

  • O3 - 0.02

  • S2 - 0.005

• The results looked really good. There is a lot of sharpening and star reduction. The detail restored to the nebula was impressive. There was a slight background pattern, but this is very subtle, and I expect Nose reduction to deal with it.

8. Linear noise reduction

I cannot use my standard Mure Denoise method as it does not work with drizzled data.

• I experimented with EZ Denoise vs. MLT

• EZ -Denoise was smoother but lost too much detail in the nebula. Based on this I decided on MLT

9. Star Artifact Problem!

• While evaluating the Denoise options, I discovered that my bright stars in Ha had a weird group of pixels sticking out from one side. See the attached image. That’s not good!

• It is not in the subs, nor is it in the normal masters.

• This must be some drizzle issue? I am not sure - but I need to fix it

The odd pixel pattern on the bright stars of my Ha image

Here is my fix - not perfect but a lot better.

• Take the local Support mask and adjust it to fits stars perfectly (using Morphological Transform operator (erosion in this case)

• Run Convolution - shape 1.5, std dev 1.9

• See the resulting image - it is not perfect, and the new image is slightly larger I was not as worried about that as I will be reducing star sizes later on. But it should make the situation more manageable.

• Has anybody run into this when using drizzle before? Enquiring minds want to know…

10. Go Nonlinear and Combine Images

• I kept this simple and used the STF mapping and Histogram tool for each image. I have experienced in the past with hand tweaked transforms using MaskedStretch and then some manual adjustment, and using LinFit to balance the layers. On complex wide-field images like this one, I have actually had better results with the STF->HT method. Your mileage may vary.

• Create a color image with ChannelCombination Tool - map images in the SHO order.

The Ha, O3, and S2 nonlinear images are ready for combination.

11. Initial Color Balance of the SHO image.

• Use SCNR to remove excess green In the image. This leaves some sky area and stars looking magenta.

• Invert the image - this gives it a color-negative look

• Magenta regions of the image now look green. Use SCNR-Green to remove the excess green

• Invert the Image again.

• Use the Curves Transform Tool to

  • Boost sats

  • Apply S-curve on neutral channel

• You now have a pretty good base image for the Hubble palette with no magenta star issues that are so common in this realm. See the images below.

Here is the initial three-colored image - just after channel combination. It’s very green!

Now I have inverted the image - note the magenta portions are now green.

Now we have the foundation for a good SHO image = we just need to boost the saturation and contrast.

SCNR has just removed the excess green. Note the Magenta in the sky background and stars.

SCNR is once again used to remove the green and now we do another invert on the image

Using the Curves Transform, I apply an S-curve to boost contrast and just pump up the saturation a bit.

12. Create Color Masks

• Blue mask - use ColorMask script with Blue chosen

• Adjust mask

  • Fix star-rings in the mask by eliminating them with the DynamicPaintBrush or CloneStamp tool

  • Run deconvolution on the mask to smooth it out

  • Boost mask contrast with HT to recover loss of intensity from Convolution operation

• Green mask - ColorMask script with Green Chosen

• Adjust mask as in the blue example above

13. Adjust colors

• Apply Blue mask

• With CT, adjust Sat and color curves to get the color effect you want on blue

• Apply the Green Mask

• With CT, adjust Sat and color curves to get the color effect you want on the Golden areas of the image

14. Do First Non-Linear DeNoise

• Use ACDNR with Chrominance set to 4.5, and Lightness set to 2.5

• Use a lightness map

• Verify lightness map coverage - adjust as needed.

• Test on several previews and adjust as needed.

• Apply on image

15. Color Rings, During the denoise work, I noticed that some Stars have color rings. This is caused by deconvolution shrinking some stars more in one color layer than another. It is best to fix this as soon as you notice it.

• Create a star mask ( I used Starnet with default values)

• Binarize it

• Run Convolution on it to soften it

• Check the fit on stars.

• Adjust the star mask with the Morphology Tool until the mask is slightly bigger than the stars.

• Apply mask and then use CT to reduce sat

16. Run DarkstructureEnhance - use default values and save the DarkStructure mask

• Did not like what it did to some star images

• Looked at the mask created and saw some star-rings. This was impacting some stars.

• I ran Convolution on the mask to get rid of the star rings

• Applied the dark mask and manually adjusted dark features with the CT tool

• Ultimately, it made the dark regions too dark, and I lost detail -so I readjusted and just added a smidgeon of darkening.

17. Run EZ-Star Reduction

18. Shift over to Photoshop

• Export image as a TIFF file - 16 bit.

• Run Photoshop, load image

19. Global Tweaks using the Camera Raw filter

• Light adjustment of Clarity

• Light adjustment of Texture

• Tweaked Curves

• Run color mixer and tweaked blues and oranges.

20. Local Tweaks - using the lasso selection tool with 130 feather settings, select key feature areas of the image and do subtle enhancements

• For selected areas - tweak with Camera Raw filter. These are very small tweaks.

  • Clarity

  • Texture

  • Curves

  • Color Mix

21. For the whole image, I did some Camera Raw Sharpening.

• This was a big mistake - I had to undo it later on.

22. Share the image with some friends for feedback

• Feedback received:

  • Blues too saturated

  • Some large stars have Yellow color.

• I also noticed that when zooming in, fine detail seemed very distorted. This was not noticeable when looking at the image as a whole, only when pixel peeping

  • This was from the PS Camera Raw Sharpen Tool! I had to go back a remove that processing step

23. Make final changes

• Go into history and undo the horrible sharpening.

• Reduce blue sat a bit

• Work on Large stars a bit

  • Make a small circular selection with a feather setting of 20

  • Pick large stars Desaturate, and used curves to tweak profiles.

• Use the StarShrink tool to shrink only the largest stars slightly.

24. Finalize image

• Add my custom watermarks

• Save image.

• Export a clean image - full res - no watermarks

• Export a full res image - with watermarks

• Export a subsampled image with higher jpeg compression for web use.

This processing took two days. You never get the image to be “perfect” at some point there are diminishing gains and you decide to stop. Perhaps during some cloud-locked winter day, I will revisit and re-process some of these images.

---

More Information

Wikipedia: NGC 7000

The LiveSky: NGC 7000

Atlas of the Universe: a nice shot of the same region (wider field) in traditional RGB. 

---

Capture Information

Light Frames

• 39 x 300 seconds, bin 1x1 @ -15C, unity gain, Astronomiks 6nm Ha Filter

• 39 x 300 seconds, bin 1x1 @ -15C, Unity gain, Astronomiks 6nm OIII Filter

• 35 x 300 seconds, bin 1x1 @ -15C, unity gain, Astronomiks 6nm SII Filter

• Total of 9.33 hours

Cal Frames

• 30 Darks at 300 seconds, bin 1x1, -15C, gain unity

• 30 Dark Flats at Flat exposure times, bin 1x1, -15C, gain unity

• Flats collected separately for each evening to account for camera rotator variances:

  • 15 Ha Flats

  • 15 OIII Flats

  • 15 2II Flats

---

Capture Hardware

• Scope: Askar FRA400 73MM F/5/5 Quintuplet Astrograph

• Guide Scope: Sharpstar 66EDPHII

• Camera: ZWO ASI1600mm-pro with ZWO Filter wheel with ZWO LRGB filter set,

• and Astronomiks 6nm Narrowband filter set

• Guide Camera: ZWO ASI290Mini

• Focus Motor: Pegasus ZWO EAF 5V

• Mount: Ioptron CEM 26

• Polar Alignment: Ipolar camera

Software

• Capture Software: PHD2 Guider, Sequence Generator Pro controller

• Image Processing: Pixinsight, Photoshop - assisted by Coffee, extensive processing indecision and second-guessing, editor regret and much swearing….. Given the problems on this image, more than the usual whining….

Click below to see the telescope platform version used for this image