rosette nebula The Rosette Nebula and Star Cluster ( NGC 2237 and 2244 ) in Monoceros ( updated )

[MikeODay]

The Rosette Nebula and Cluster ( NGC 2237 and 2244 ) in the constellation Monoceros

edit:  updated 30th Dec with improved colour balance and slightly increased brightness ...

 

......

original:

( please click / tap on image to see larger and sharper )

Still a work-in-progress really... with only 10 x 4min exposures for the main 'lights' before the clouds came over.  I will try to add some more data when the moon has gone

I am still experimenting with how to get the best out of the D7500.  With the very warm nights ( low to mid 20s all night ) the 'warm pixels' are very noticeable so I reverted to my old practice of in-camera dark subtraction.  This worked quite well and produced a nice smooth noise floor in the integrated images - albeit at the expense of more exposures.

.................

Identification:

The Rosette Nebula ( NGC 2237 ) is a large, circular emission nebula in the constellation Monoceros.  It surrounds a cluster of hot, young stars known as the Rosette Cluster ( NGC 2244 ). ( SkySafari )

NGC 2237, 2244
Caldwell 49, 50

North is up.

..................

Capture Details:

Telescope: Orion Optics CT12 Newtonian ( mirror 300mm, fl 1200mm, f4 ).
Corrector: ASA 2" Coma Corrector Quattro 1.175x.
Effective Focal Length / Aperture : 1400mm f4.7

Mount: Skywatcher EQ8
Guiding: TSOAG9 Off-Axis-Guider, Starlight Xpress Lodestar X2, PHD2 

Camera:
Nikon D7500 (unmodified) (sensor 23.5 x 15.7mm, 5568x3712 @ 4.196um pixels)

Location:
Blue Mountains, Australia 
Moderate light pollution ( pale green zone on darksitefinder.com map )

Capture ( 23 Dec 2017 )
9 sets of sub-images with exposure duration for each set doubling ( 1s to 240s ) all at ISO400.
10 x 240s + 5 each @ 1s to 120s
imaged ~ +/- 1.5hrs either side of meridian
maximum altitude ~ 51.3 deg above north horizon

Processing ( Pixinsight )
Calibration: master bias, master flat and in-camera dark subtraction
Integration in 9 sets
HDR combination 

Image Plate Solution
===================================
Resolution ........ 0.633 arcsec/px ( full size image )
Rotation .......... 0.181 deg
Focal ............. 1367.90 mm
Pixel size ........ 4.20 um
Field of view ..... 58' 59.4" x 39' 15.0"
Image center ...... RA: 06 31 55.638  Dec: +04 56 30.84
===================================

[Datalord]

9 hours ago, MikeODay said:

9 sets of sub-images with exposure duration for each set doubling ( 1s to 240s ) 

[Datalord]

I'm too stupid to make the quote system work in edited form.

Anyway, why do the different exposure durations?

[MikeODay]

2 hours ago, Datalord said:

I'm too stupid to make the quote system work in edited form.

Anyway, why do the different exposure durations?

To try to keep colour in the bright stars I produce an HDR image with over 99% of the image coming from the main long exposure set of images and the rest ( ie stars and burnt out centres of galaxies, etc ) being replaced by the range of shorter exposures.  I have found that decreasing the exposure of each subsequent set by 1 ‘stop’ overcomes the ‘onion’ effect that I was seeing in the stars when the exposure difference was much larger.

For example, without combining a range of exposures the centres of these bright stars would just be burnt out round disks.

[Datalord]

Very interesting. But won't this introduce a lot of noise?

[MikeODay]

7 hours ago, Datalord said:

Very interesting. But won't this introduce a lot of noise?

No not really - the combination is performed using masks.  Only the saturated parts of the image are replaced with the unsaturated portions from the less exposed images.

[MikeODay]

Processing updated 30th Dec - improved colour balance and somewhat increased brightness.

The Rosette Nebula and Star Cluster ( NGC 2237 and 2244 ) ( please click / tap image to see larger )

And a black and white version ...

And inverted ...

[Datalord]

I am thoroughly impressed with your stars! I'm trying out the technique right now, just for testing the technique.

If I would have anything constructive to say about the picture, it is that I think it is a bit too noisy for my liking. Whether it is a tiny bit stretched too far or you could smoothen the nebulosity a bit more depends on your process. But that is nitpicking. A very beautiful image.

[MikeODay]

19 hours ago, Datalord said:

I am thoroughly impressed with your stars! I'm trying out the technique right now, just for testing the technique.

If I would have anything constructive to say about the picture, it is that I think it is a bit too noisy for my liking. Whether it is a tiny bit stretched too far or you could smoothen the nebulosity a bit more depends on your process. But that is nitpicking. A very beautiful image.

Thanks for that.

Yes it is a tad noisy.  Three main causes - the Rosette Nebula is very faint against the bright skies I have here and I really need more data; the night this was taken stayed warm ( around 20deg C ) adding significant thermal noise to my uncooled camera; and it was taken with my new Nikon D7500 which I have just established has a much bigger problem with thermal noise than my old D5300.  I did try more aggressive noise reduction / smoothing ( thanks for the suggestion ) but I was not happy with the loss of detail and definition, particularly in the dark nebula.  If the sky clears over the next day or so I will try to get more data with my old D5300 and see if I can improve it.

Cheers

Mike

[Datalord]

56 minutes ago, MikeODay said:

Thanks for that.

Yes it is a tad noisy.  Three main causes - the Rosette Nebula is very faint against the bright skies I have here and I really need more data; the night this was taken stayed warm ( around 20deg C ) adding significant thermal noise to my uncooled camera; and it was taken with my new Nikon D7500 which I have just established has a much bigger problem with thermal noise than my old D5300.  I did try more aggressive noise reduction / smoothing ( thanks for the suggestion ) but I was not happy with the loss of detail and definition, particularly in the dark nebula.  

Cheers

Mike

If you use PS (or probably any other tool), you can copy two layers from your base. The first layer you reduce noice all over, in the second you erase the layer on the nebulosity to reveal the noise-reduced nebulosity, but keeping the sharpness of the structures. It's a bit of work, but has helped me since I learned that little trick.

[MikeODay]

7 minutes ago, Datalord said:

If you use PS (or probably any other tool), you can copy two layers from your base. The first layer you reduce noice all over, in the second you erase the layer on the nebulosity to reveal the noise-reduced nebulosity, but keeping the sharpness of the structures. It's a bit of work, but has helped me since I learned that little trick.

Thanks for the tip, I will look into it and have a go.

Cheers

Mike

[Datalord]

On 12/24/2017 at 21:43, MikeODay said:

To try to keep colour in the bright stars I produce an HDR image with over 99% of the image coming from the main long exposure set of images and the rest ( ie stars and burnt out centres of galaxies, etc ) being replaced by the range of shorter exposures.  

 

Any chance you have seen a tutorial for doing this somewhere? I can't figure out how you do this.

[MikeODay]

19 hours ago, Datalord said:

Any chance you have seen a tutorial for doing this somewhere? I can't figure out how you do this.

I do not recall seeing a tutorial.

I use Pixinsight to produce HDR images.  I would be happy to run through my steps but I am not sure how relevant it would be to other applications.  Do you use Pixinsight?

Anyway, basic steps ( assuming image sets with doubling of exposure time ):

1. Calibrate, debayer, align all frames

2. Separate into different exposure values and integrate each set separately

3. Check the maximum values in each channel of all integrated images
- find the image that has the maximum exposure duration such that no channel is clipped.  This becomes the miniumum exposure image and all images with less exposure can be discarded.
- ensure that all channels in all images with longer exposure than the minimum exposure image are clipped.  
  That is, in Pixinsight;  $T = $T / min ( max ( $T[0] ), max ( $T [1] ), max ($T[2]) )
  Or in PS, adust white point to the left such that all channels are ( just ) clipped.

4. Produce the HDR image before stretching:

In Pixinsight, one can use the HDR combination function to combine all images to produce an HDR image.  Or, take the masks from this step and combine them manually using PixelMath { this is what I do and I can dig out the PixelMath formula if you are interested }.

If you use PS you might try building up a stack of layers ( longest exposure at the bottom ) with masks inserted between them to only allow the new non-clipped portions ( suitably feathered ) to pass upwards in the stack.  That is, using selection masks, recursively replace the clipped portion of the lower image in the stack with the corresponding non-clipped portion from the next higher image, etc.  

Having said that, I am not sure how well PS version above will work given that this process will produce a non-linear image.  That is, the exposure difference between each subsequent exposure is 2 times and as such, the pixel values in the area immediately around each masked portion will involve a significant jump in value ( albeit that the selections should be feathered ).   When I produce the HDR image manually in Pixinsight, I recursively divide by 2 the built up image before I add in the new portion.  This maintains linearity.  I don’t know how you would do the same in PS.
 

I hope that helped ...

Cheers

Mike

 

 

[MikeODay]

21 hours ago, Datalord said:

Any chance you have seen a tutorial for doing this somewhere? I can't figure out how you do this.

I did a quick search and came up with the following:

http://astrophotolinks.com/category/hdr/

http://www.deepskycolors.com/archive/2011/01/16/hdr-Composition-for-astronomical-image.html

http://www.astropix.com/html/j_digit/ps_hdr.html

http://www.lightvortexastronomy.com/tutorial-producing-an-hdr-image.html

 

[Datalord]

Thanks for these!

I think I will have to put this on the list of "future learning". I don't have or use PI and the more I look at pictures and read about it, this seems to be a Newtonian thing. I tried to take a series of pictures on the Pollux region and I just couldn't see that kind of spike you have. But, I think I will be able to use the HDR process for nebula cores as well, so I appreciate it.

[MikeODay]

On 10/01/2018 at 06:02, Datalord said:

Thanks for these!

I think I will have to put this on the list of "future learning". I don't have or use PI and the more I look at pictures and read about it, this seems to be a Newtonian thing. I tried to take a series of pictures on the Pollux region and I just couldn't see that kind of spike you have. But, I think I will be able to use the HDR process for nebula cores as well, so I appreciate it.

Yes the spikes are unrelated to HDR processing.  They are due to diffraction caused by the ‘spider vanes’ that hold the secondary mirror in place.  All the HDR processing is giving me is an increase in the dynamic range from somewhere near 18bits ( after integration of the maximum exposure images ) to around 30 bits ( for the HDR image ).

[ollypenrice]

I think the image is lovely. Some of the processing strikes me as complicated but it's the result we look at - with pleasure in my case!

Olly

[Datalord]

1 hour ago, MikeODay said:

Yes the spikes are unrelated to HDR processing.  They are due to diffraction caused by the ‘spider vanes’ that hold the secondary mirror in place.  All the HDR processing is giving me is an increase in the dynamic range from somewhere near 18bits ( after integration of the dsmaximum exposure images ) to around 30 bits ( for the HDR image ).

Haha, I live and learn. No wonder I had a hard time figuring out what was going on. 

This is what goes on in my scope, which imho is so much less desirable. I do think your HDR technique will help me eventually to control the bloat of the stars, so I'm not giving up. Thanks again.

[Shibby]

That's a very nice looking image, especially considering the short integration time. Some targets actually look nicer when a DSLR is not modified, as the excess of Hα would otherwise wash colours out with red.

Avoid using in-camera dark subtraction. You'll get far less noise if you stack your darks separately to create a master calibration image.

[MikeODay]

23 hours ago, ollypenrice said:

I think the image is lovely. Some of the processing strikes me as complicated but it's the result we look at - with pleasure in my case!

Olly

Thanks Olly.  Yes it is does seem a little over complicated, particularly when written down .  I am still working on how to optimise the workflow.

[MikeODay]

8 hours ago, Shibby said:

That's a very nice looking image, especially considering the short integration time. Some targets actually look nicer when a DSLR is not modified, as the excess of Hα would otherwise wash colours out with red.

Avoid using in-camera dark subtraction. You'll get far less noise if you stack your darks separately to create a master calibration image.

Thanks Lewis.  I believe you are right with respect to in-camera dark subtraction versus using a master dark, at least when the master dark is closely matched to the noise in the set of images.  The challenge it to achieve that match when the temperature varies widely during a session and ones camera is un-cooled.  I plan to go back to my old D5300 which has far less of an issue with thermal noise so on cold nights I can ignore the small amount of thermal noise ( which is overwhelmed by sky noise ) and when it is hot, use a master dark and rely on Pixinsight to scale it to approximately to match the noise in the images.

Cheers

Mike

[Datalord]

1 hour ago, MikeODay said:

Thanks Lewis.  I believe you are right with respect to in-camera dark subtraction versus using a master dark, at least when the master dark is closely matched to the noise in the set of images.  The challenge it to achieve that match when the temperature varies widely during a session and ones camera is un-cooled.  I plan to go back to my old D5300 which has far less of an issue with thermal noise so on cold nights I can ignore the small amount of thermal noise ( which is overwhelmed by sky noise ) and when it is hot, use a master dark and rely on Pixinsight to scale it to approximately to match the noise in the images.

Cheers

Mike

I just noticed the gear you have and can't help wondering why you don't buy a proper cooled camera? You seem to have a very potent setup, but the camera kinda sticks out.

[MikeODay]

2 hours ago, Datalord said:

I just noticed the gear you have and can't help wondering why you don't buy a proper cooled camera? You seem to have a very potent setup, but the camera kinda sticks out.

Well, it’s a question of bang for the buck.

I have built up my equipment over time, as funds allow,  by choosing items that I thought would make the most difference to the final images.  For me the priority has been on the OTA and the mount.

I have understood that cooling is useful in situations where the thermal noise is apparent and significant.  That is, the ambient temperature in high, the skies are very dark and/or the camera suffers from relatively high thermal noise.  With the D5300, which is a low noise camera, and my moderately light polluted skies, the thermal noise has rarely been a significant factor and in fact I stoped using darks with this camers.  Note that this is not the case with the D7500, where on warm nights in particular the thermal noise dominates the sky noise.  ( As an aside, the D7500 has been a disappointment to me; I assumed that, as it is a new model with lower read noise, the D7500 would have lower overall noise than the D5300 but this not the case. )

My plan now is to sell my D7500 and go back to using my D5300.  I will keep an eye on the results people are getting with the latest Chinese CMOS colour astro cameras ( in particular the QHY247C - which has the same sensor as the D5300 )  and, when funds allow, consider buying one for the expected small improvement in overall noise and, perhaps more importantly, for the less aggressive filtering of IR light.

[Datalord]

2 minutes ago, MikeODay said:

My plan now is to sell my D7500 and go back to using my D5300.  I will keep an eye on the results people are getting with the latest Chinese CMOS colour astro cameras ( in particular the QHY247C - which has the same sensor as the D5300 )  and, when funds allow, consider buying one for the expected small improvement in overall noise and, perhaps more importantly, for the less aggressive filtering of IR light.

Coincidence that I have a QHY247C and I have no regrets with it. Compared to the Canon 6D, which is also an excellent camera, I have much less noise. If there is something you want to see with this in a single frame, I'll be happy to take a shot next time I have clear skies.

[MikeODay]

6 hours ago, Datalord said:

Coincidence that I have a QHY247C and I have no regrets with it. Compared to the Canon 6D, which is also an excellent camera, I have much less noise. If there is something you want to see with this in a single frame, I'll be happy to take a shot next time I have clear skies.

That is really very kind of you.  

What would be ideal, if you already have one, would be a master dark based on subs taken with parameters close to the ones I used to capture the dark frames I have for the D5300 and D7500.

That is,

- 16bit raw 

- gain: 2200 ( ie. Unity gain where 1e=1ADU , which according to the QHY website is 2200 for the QHY247c )

- offset: 2400 ( based on 16 bit scale )

- exposure duration: 240 seconds

- no calibration or other processing

If you don’t already have such a master dark then, if it is not too much trouble, a single dark taken with those parameters would be wonderful.

I could compare this file with the darks / master darks I have for the D5300 and D7500 ( refer to link below ), in particular focusing on an examination of amp glow, noise level and any pattern in the noise ( in the case of a master dark ).

If you have dropbox or similiar then the full raw file would be good.  If you have limited bandwidth then perhaps I could suggest a integer based decimation ( say div by 4 ) to produce a smaller file.

Thanks and much appreciated

Mike