New Method for Star De-Emphasis by Adam Block

[ngc1535]

Hi all,

Recently I published my method for de-emphasizing stars. This method does not use an erosion filter and so avoids the pitfalls/artifacts of the method.
I demonstrated the technique during my guest appearance on the Astro Imaging Channel last Sunday April 20th 2019.
In addition, because it is popular in some circles, I also went through the exercise of creating a written version of the technique (though I still think the video methods are always best).

Please visit the free material section of my website here:

https://www.adamblockstudios.com/categories/Free_Content

You will find both the embedded YouTube video as well as a link to the written version (below):

https://www.adamblockstudios.com/articles/star_demph

On a meta-note concerning PI- I showed this technique during my invited talk and workshop at NEAIC a few weeks ago. It was titled "Innovations in PixInsight" and the idea behind the talk was to highlight not only what PI does- but what it can be made to do. PixInsight by design permits a user to develop methods and tools that may not exist due to its open ended implementation. Although I have seen many talks on how to use PixInsight; but I was striving to show yet a greater level of abstraction- one that is very powerful as I hope my innovation above shows.

Even if the technique isn't your cup of tea- you might find the sub-techniques I used of interest in and of themselves. Although teaching PI is part of my "business"- it is also important to give back to a community that has always been extremely supportive of my efforts.

Someone who watched the video published their results (and was happy!) on the Cloudy Nights Forum.
https://www.cloudynights.com/topic/658570-same-data-different-processing-the-wizard-nebula/#entry9309605

It is a subtle processing effect... especially at this scale- but note how the nebulosity is easier to see with the de-emphasis of the stars in my example image**,



Sincerely,
Adam Block

** This image example was kindly provided to me by Terry Robison. It is always best to demonstrate your techniques on other people's images.

[Juan Conejero]

Hi Adam,

Congratulations on a nice and clever method of star reduction. I like the way you apply multiscale analysis. A good example of the outstanding usefulness of these algorithms, especially MMT to isolate structures. The use of artificial noise to generate seamless transitions is also an interesting algorithmic idea. I am implementing a related technique (randomization) for seamless mosaic generation. An effective and highly controllable technique, very recommendable.

[llpastro]

Adam, I tried to copy your instructions from your link to the article but it appears your website is copy protected.  Any way to print the instructions for use offline?

Larry

[ngc1535]

Adam, I tried to copy your instructions from your link to the article but it appears your website is copy protected.  Any way to print the instructions for use offline?

Larry

Hi Larry,

Right click is disabled on my site (which I assume is how you tried to copy the information). However, if you use your keyboard to get the contents into your clipboard, it will work.
In Windows CRTL+C . I was able to copy and paste into a Word document OK (it copied everything... font, background, pictures... lol).
-adam

[msmythers]

Larry

I right clicked on the side of the article and was able to save it as a PDF. I'm using Opera as my browser. I just checked and I was able to view the saved pdf in my other browsers.



Mike

[llpastro]

Thanks Mike, that worked.
Larry

[STEVE333]

Very nice technique Adam - You are a creative wizard.

I gave it a go on an image of the East/West Veil. I didn't fully process the image. Just wanted to see how your technique worked. I processed the image up through Linear processing with a light MLT noise reduction. I then combined a HT stretch with an Arcsinh stretch (50/50) and that is all the processing that was done.

Your technique reduced the brightness of the smaller stars beautifully. I need to go back and "mask out" the brighter stars to prevent artifacts (any suggestions gratefully suggested). A side-by-side comparison of a portion of the image is shown in the attachment. The image on the left is from the original image and the image on the right shows the results of using your technique (only one iteration).

I had tried another technique for removing stars. It removed the stars beautifully, but, it also blurred out some of the nebula details. Your technique left all the nebula details "untouched".

Well done, and, thanks for sharing.

Steve

[STEVE333]

Just FYI - I subtracted a modified RangeSelection mask from the Halos mask to protect the larger stars. That removed most of the artifacts from the "dimmed stars" result.

Really like your process.

Thanks again.

Steve

[Solidarity]

Hello Adam,

very good Method to dim the stars.

In my Image the effect is very strong and my question is how to have smaller amounts of dimming. Is it caused by a "wrong" mask?

In the ataatched image of M42 with DSLR you will see the effect on the image which is not yet processed to the end.

Greetings from germany

Karsten

[ngc1535]

Hi Karsten,

Sorry for the late reply- I was away from the computer for a few days.
I suspect your mask is fine. The issue is that when the stars are very small (like in undersampled images)- the mask doesn't make halos that protect the cores of stars...the mask pretty much covers these tiny stars. You can modify the halo mask by forcing there to be black holes in the tiny stars- there are a number of ways to do it- but I suspect this is the primary issue. If I am correct- you could upsample the image (make it larger) or perhaps apply the star de-emphasis to drizzled images (which are larger). This will give you a bit more control with the small stars so that you are not making a starless image (which is a different effect).

-adam

[Solidarity]

Hello Adam,

thank you for your reply. I will try to modify the Halo Mask - I have to try it. At the moment I don't know how but I will think about it.

I used your process on the drizzeld image and will do it a second time.

Greetings

Karsten

[jtalbot]

A truly creative technique Adam.  I've enormously enjoyed your video series PixInsight Horizons.  Even though I've been a user for many years I learn something new every time from your video tutorials as they show ways to handle data in ways I never would have thought.  Thanks for sharing this to the rest of the community.

Jon

[lucam]

I really like this method. Thank you, Adam for making the documentation available.

I have used it on my image of the Iris Nebula (NGC 7023) to tame some of the medium size stars. I found that for this image in order to match the chromaticity of the noise, it was easier to add noise to the luminance of the pixel substitution image (Sub_pix) and then channel combination the noise-added luminance into the color Sub_pix image, instead of trying to match hue and saturation of the noisy patches to the background.

Animated GIF of a crop of the image with before/after star de-emphasis attached. I applied star de-emphasis right after LRGB combination, before HDRMT, saturation, contrast and sharpness adjustments.

The full size before/after image files are posted on my Astrobin page in panels D and E: https://astrob.in/419457/F/.

[ngc1535]

Great jo!..and very good suggestion concerning the noise in the substitution image.

I note that not all stars, especially near the nebula, were affected. This may have been intentional or a by-product of how you created the star mask.

I just recently created a tutorial that demonstrates how to use ImageSolve in combination with Harmut Bornemann's MaskGen to create a star mask from a star catalog.
This method does an end run around structure orientated star masks that are sometimes commingled with significant structure of other objects.

-adam

[lucam]

Great jo!..and very good suggestion concerning the noise in the substitution image.

I note that not all stars, especially near the nebula, were affected. This may have been intentional or a by-product of how you created the star mask.

I just recently created a tutorial that demonstrates how to use ImageSolve in combination with Harmut Bornemann's MaskGen to create a star mask from a star catalog.
This method does an end run around structure orientated star masks that are sometimes commingled with significant structure of other objects.

-adam

Thank you, Adam.

I purposely excluded stars in the blue nebula. They are too different from the stars buried in the dust and would have required a separate noise and color matching step. They are also sparse and not that big, so I didn't find them to detract from the image. I have attempted creating a star mask with MaskGen and it worked ok. Some of the stars were missed altogether (not just little ones, kind of randomly) and it wasn't obvious to me how to match the star sizes in my image. However it was impressively close.

--Luca