[RCC] M16 Eagle Nebula - Pillars of Creation Requests for constructive critique · Andreas Eleftheriou · ... · 17 · 674 · 4

andreas1969 6.02
...
Dear all,
I would like to submit this image latest revision https://www.astrobin.com/full/407761/D/?nc=andreas1969 for Constructive Critique:
 - The amount of small detail structures one is able to collect using a refractor VS a bigger reflector
-Post processing techniques and amount of deconvolution, noise reduction and sharpening, while maintaining the natural look of the image/target

In general, I was surprised by the rich small-scale details I was able to reveal from the Ha data (4hours) collected recently with the Esprit150 and the QSI690 camera and applied to the older colour image as L. This rich and smooth data allowed me to apply deconvolution (PixInsight)  very successfully at the very first stages of the post processing procedure. I have used Topaz Labs DeNoise AI while masking the stars in the final stage of the process.

Any comment regarding the data acquisition quality and method, as well as the post processing procedure and technique, are welcome and I will be happy to share my comments as well.

Best regards and clear skies,
Andreas
Like
carastro 8.04
...
Looks a great image to me, nice detail and good processing.

My only suggestion is to add RGB stars to it as " icing on the cake" as we say in the UK.

Carole
Edited ...
Like
andreas1969 6.02
...
Thanks Carole! I will definitely try that. The only issue is that Ha stars are much smaller than the RGB stars so I need to find a good balance on star color and star size :-)
A.
Like
olaskarpen 1.20
...
Nice editing, but I would definitely remove all the green in the image.(My personal opinion)

Skarpen
Like
matherneconnor 0.90
...
Andreas Eleftheriou:
- The amount of small detail structures one is able to collect using a refractor VS a bigger reflector-Post processing techniques and amount of deconvolution, noise reduction and sharpening, while maintaining the natural look of the image/target

Always love threads like this, only the brave put their work up for others to critique! In an effort to improve the image, I would suggest that there is too much noise reduction in the dust lanes. A lot of the structures here are virtually noise free, but at the same time, they lost a lot of detail relative to the other parts of the image.

Additionally, the stars are devoid of any colors. Even without taking RGB data, you can still leave some awesome color variations within the stars. See below



Lastly, I think the bottom portions of the image that are a bit darker should still be colored. There is still a ton of Ha in those areas, no need to kill it off. While it may give the Pillars more of a "in a cave" appearance, I think the extra color like a dark gold or red would be a nice addition.

Hope these are insightful!
Connor
Like
KuriousGeorge 1.20
...
Really nice detail. I happened to just start this object. Here's an overprocessed sample in a CDK24 to help me check the Ha sub quality...

https://www.astrobin.com/l834me/F/

This is 18 x 5 minutes Ha with subs between 1.93" and 2.15" FWHM.
Like
andreas1969 6.02
...
Ola Skarpen SkyEyE:
Nice editing, but I would definitely remove all the green in the image.(My personal opinion)

Skarpen

Thanks Skarpen. I will give it a try!
Andreas
Edited ...
Like
andreas1969 6.02
...
Connor Matherne:
Andreas Eleftheriou:
- The amount of small detail structures one is able to collect using a refractor VS a bigger reflector-Post processing techniques and amount of deconvolution, noise reduction and sharpening, while maintaining the natural look of the image/target

Always love threads like this, only the brave put their work up for others to critique! In an effort to improve the image, I would suggest that there is too much noise reduction in the dust lanes. A lot of the structures here are virtually noise free, but at the same time, they lost a lot of detail relative to the other parts of the image.

Additionally, the stars are devoid of any colors. Even without taking RGB data, you can still leave some awesome color variations within the stars. See below



Lastly, I think the bottom portions of the image that are a bit darker should still be colored. There is still a ton of Ha in those areas, no need to kill it off. While it may give the Pillars more of a "in a cave" appearance, I think the extra color like a dark gold or red would be a nice addition.

Hope these are insightful!
Connor


Thanks Connor, I appreciate your comments and the detailed suggestions. You are right about the star colours, I was focused on revealing the delicate structures on the pillars and the dust lanes. I tried not to loose too much details when applying noise reduction, I might try a more subtle approach and check the results :-). Nice image and colours by the way!
Andreas
Like
andreas1969 6.02
...
Really nice detail. I happened to just start this object. Here's an overprocessed sample in a CDK24 to help me check the Ha sub quality...

https://www.astrobin.com/l834me/F/

This is 18 x 5 minutes Ha with subs between 1.93" and 2.15" FWHM.

Thanks George. It is always a challenge to reveal the structures as seen in the iconic HST image! with the CDK24 I am sure you will get much more details! :-)
Like
KuriousGeorge 1.20
...
Andreas Eleftheriou:
Thanks George. It is always a challenge to reveal the structures as seen in the iconic HST image! with the CDK24 I am sure you will get much more details! :-)


Your detail is among the best I've seen! May I ask what was the FWHM on your Ha subs and what's your altitude above sea level?

Also, may I ask what sharpening technique you used?
Like
Bobinius 9.01
...
Hi Andreas,

First let me say that the pillars are magnificent, the central part of the image is amazing. I agree with Connor, I think the main improvement would be to reveal more details and structure in the periphery. I looked at the mouseover image comparison and I think you lost depth and structure separation in the periphery around the pillars with the final processing (while improving the central part of the image). It is a bit too stretched and the noise reduction decreased local contrast. So maybe even without new data (the result is excellent for only 7.5h) you can improve the image and make it more balanced. My two cents : ).

Best regards

Bogdan
Like
andreas1969 6.02
...
Andreas Eleftheriou:
Thanks George. It is always a challenge to reveal the structures as seen in the iconic HST image! with the CDK24 I am sure you will get much more details! :-)
Your detail is among the best I've seen! May I ask what was the FWHM on your Ha subs and what's your altitude above sea level?

Also, may I ask what sharpening technique you used?


FWHM of the Ha sub is 1.86" or 2.58 pixels. My Guiding RMS was 0.6". My altitude above sea level is just 20m :-) (I am 800m from the beach).
Regarding sharpening, I used Deconvolution on the brightest areas while in linear stage. After stretch, I used HDRMTransformation and mild LocalHistogram Equilisation, using a simple L mask. At the end, I used TopazLabs DeNoise AI while masking the stars.  I believe Deconvolution (on which I spent some time on finetuning the mask and the settings) have provided a strong basis for the rest of the post processing procedures.

Regarding subframes, I used PI Frame Selection tool and I selected/stacked around 80% of the subs.

I am happy to share and compare Pixinsight workflow techniques with the rest of the community 
Andreas
Like
andreas1969 6.02
...
Bogdan Borz:
Hi Andreas,First let me say that the pillars are magnificent, the central part of the image is amazing. I agree with Connor, I think the main improvement would be to reveal more details and structure in the periphery. I looked at the mouseover image comparison and I think you lost depth and structure separation in the periphery around the pillars with the final processing (while improving the central part of the image). It is a bit too stretched and the noise reduction decreased local contrast. So maybe even without new data (the result is excellent for only 7.5h) you can improve the image and make it more balanced. My two cents : ).

Best regards

Bogdan


Thanks Bogdan, Your two cents are  much appreciated :-) The newly acquired Ha Data (Esprit150)  is much richer in small details than the old (10"RC). I will agree that the noise reduction could be milder. I will try to collect O3 data awith the Esprit150 the next few days and I will process them again less aggressively  in a new HOO palette instead of SHO :-)

Thanks again and feel free to provide any additional comments and suggestions. I have learned a lot the last few years since I joined Astrobin from this community :-)

Andreas
Like
KuriousGeorge 1.20
...
Andreas Eleftheriou:
FWHM of the Ha sub is 1.86" or 2.58 pixels. My Guiding RMS was 0.6". My altitude above sea level is just 20m :-) (I am 800m from the beach).Regarding sharpening, I used Deconvolution on the brightest areas while in linear stage. After stretch, I used HDRMTransformation and mild LocalHistogram Equilisation, using a simple L mask. At the end, I used TopazLabs DeNoise AI while masking the stars.  I believe Deconvolution (on which I spent some time on finetuning the mask and the settings) have provided a strong basis for the rest of the post processing procedures.

Regarding subframes, I used PI Frame Selection tool and I selected/stacked around 80% of the subs.

I am happy to share and compare Pixinsight workflow techniques with the rest of the community 
Andreas


Great info Andreas! You have very steady skies and great tracking.

I also use Deconvolution and HDR/LHE frequently – and this workflow appears to work very well for this object. You may also wish to try Shadows/Highlights in Photoshop. That function is very similar to PI HDR/LHE.

I haven't integrated TopazLabs DeNoise AI yet. Do you find that tool has the potential to add "detail" that may not be in the original image? For example, might their AI be smart enough to find similar images on the web and use that for very intelligent sharpening and noise reduction? That would be most interesting. 
Edited ...
Like
andreas1969 6.02
...
:-)))
that would be very interesting indeed!
DeNoise AI is a very good tool for astrophotography in noise reduction and detail sharpening, but it can create strange artifacts around small stars. I try to blend the Denoise image with the original with the stars masked in order to avoid this.
Like
andreas1969 6.02
...
Ola Skarpen SkyEyE:
Nice editing, but I would definitely remove all the green in the image.(My personal opinion)Skarpen


Green cast removed. Thanks Skarpen :-)
https://www.astrobin.com/407761/E/
Like
olaskarpen 1.20
...
Andreas Eleftheriou:
Ola Skarpen SkyEyE:
Nice editing, but I would definitely remove all the green in the image.(My personal opinion)Skarpen
Green cast removed. Thanks Skarpen :-)
https://www.astrobin.com/407761/E/

👍👍
Like
andreas1969 6.02
...
Thanks everyone for your comments and suggestions. I have reprocessed, cropped, re-framed and reposted the image at https://www.astrobin.com/axadyf/0/
A cropped Pillars image can also be found at https://www.astrobin.com/axadyf/B/?nc=user

Thanks again!
CS
Andreas



Edited ...
Like
 
Register or login to create to post a reply.