Jump to content

Narrowband

Horsehead HaRgb


Allinthehead

Recommended Posts

26 minutes ago, Allinthehead said:

Thanks Brendan, could you point me to the discussions you mention? 

I was not involved in any discussions - so cannot point to them, sorry. I believe that this first raised its head when a guy named Wassam Ayub (I may have spelt his name wrong) got a NASA APOD for an image that was so detailed....with the help of the Topaz software. I myself thought that image was brilliant (like yours here) but I can't help wondering if I am ready to accept this tool just yet. Looking back on how astrophotography has delveloped and improved over the years one must look at this software as just another tool. Being a tool that (as far as I can see) has such a huge impact on the finished image and as we are in the early days of it being used, I would just like to know when it has impacted on the image that I am looking at. Eventually, most if not all astro photographers may use this or similar and we wont bat an eyelid......but I doubt that it will be acceptable to professional astronomers if it does actually fill in the gaps where the noise was taken out.

Like you, I would love to see a good evaluation of what it actually does. For my small trial - I did not get too excited about it....but that does not mean that at some time in the future I will again decline to buy it and use it.

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

36 minutes ago, Allinthehead said:

This is the data that APP spat out when stacked. I'll throw up what it did to B150 later.

Wow, you really get a feel for the noise reduction on the final image.  In comparison the final image looks almost painted!

Echoing my earlier Astrobin comment, fantastic image Richard 👍  Detail and colour to pore over for ages!

  • Like 1
Link to comment
Share on other sites

It has certainly done more than just denoising, there appears to be some selective sharpening too. Whether or not that detail is really there or is a product of the AI aspect of the software is another question.

image.png.134527d47f0e2cb48dc91dace04af8c3.png

Regardless of the answer, the final result makes for a truly stunning image.

Cheers

  • Like 1
Link to comment
Share on other sites

8 minutes ago, Kinch said:

Interesting.....is (for me anyway) doing more than de-noise and I think this explains how final images have that OOOOHH factor. In the end a personal choice.....

 

 

Before & After.JPG

That really shows the impact. I should've added that there're two sliders. One for noise and one for sharpening. This was done with both sliders at about 50%. The slider for sharpening is the one that seems to adding detail. I need to look into it more because I've no interest in using a tool like this if it's adding something that isn't in the data.

Link to comment
Share on other sites

1 minute ago, Spongey said:

It has certainly done more than just denoising, there appears to be some selective sharpening too. Whether or not that detail is really there or is a product of the AI aspect of the software is another question.

image.png.134527d47f0e2cb48dc91dace04af8c3.png

Regardless of the answer, the final result makes for a truly stunning image.

Cheers

Indeed, I've just clarified that the tool does have a sharpening aspect.

Link to comment
Share on other sites

Just now, Allinthehead said:

Indeed, I've just clarified that the tool does have a sharpening aspect.

I'd be interested in the output of the tool with only denoise functionality, if you could do such a test?

Link to comment
Share on other sites

22 minutes ago, Allinthehead said:

I need to look into it more because I've no interest in using a tool like this if it's adding something that isn't in the data.

Is it adding something or just able to display what perhaps the optics/resolution/conditions weren't quite able to?  From the flame nebula crop, I just see genuine detail being brought out - magically almost! - rather than anything being added/manipulated.... unless that's the issue? (resolving detail that perhaps wasn't quite there)?   

Edited by geeklee
Removed mildly facetious comment! :)
  • Like 1
Link to comment
Share on other sites

5 minutes ago, Allinthehead said:

I've no interest in using a tool like this if it's adding something that isn't in the data.

But that is what is changing.....you will find in time that if you are the only one not using such a tool, then yours will be the worst looking image regardless of the time & effort you put in getting the data and putting your years of processing back into the final image.

For me....we cannot stop this ....but until is is more common and generally used, I would just love people to let me know that Topaz AI is in the equation somewhere.  

  • Like 1
Link to comment
Share on other sites

Just now, Kinch said:

But that is what is changing.....you will find in time that if you are the only one not using such a tool, then yours will be the worst looking image regardless of the time & effort you put in getting the data and putting your years of processing back into the final image.

For me....we cannot stop this ....but until is is more common and generally used, I would just love people to let me know that Topaz AI is in the equation somewhere.  

Agreed, providing the author is up front and honest with whatever processes or steps (s)he has taken to produce the image, then I have no problem.

If people were claiming that some structure was present when in fact it is an artefact of software, then I would take issue. However, I don't think many people would do that, and at the end of the day, we're (or at least I am) in the business of taking pretty pictures, not performing scientific analysis on structural detail.

  • Like 1
Link to comment
Share on other sites

18 minutes ago, Allinthehead said:

This was done with both sliders at about 50%

I have found with my own data, that I can push the sliders certainly above 10.  I wonder does it depend on the imaging scale and resolution of the camera, and quality of the data to begin with.  It doesnt look to me like it is adding anything, certainly nothing more than any of the other PS processes do.

  • Like 1
Link to comment
Share on other sites

3 minutes ago, Spongey said:

and at the end of the day, we're (or at least I am) in the business of taking pretty pictures, not performing scientific analysis on structural detail.

That is not exactly my thinking on what I am doing. In my processing I try to pull the detail out that my scope and camera have captured. I will never get the detail with a 5" refractor that someone with a 17" Planewave will get....so I know there are limits.  Overall it is that challenge that keeps me going. If the software too easily removes/lessens the challenge that I have with my gear.....then I think I would lose interest. It is not a matter of just making pretty pictures - we can do that perhaps anytime - and indeed some people do it with pretty colours.

We are all in this for different reasons - and that is precisely why some people will buy this software and others won't. Neither approach is right or wrong - but it would be nice to know (when we are looking at someone else's image) because it appears to make such a fundamental difference to the final image.

  • Like 2
Link to comment
Share on other sites

30 minutes ago, geeklee said:

Is it adding something or just able to display what perhaps the optics/resolution/conditions weren't quite able to?  From the flame nebula crop, I just see genuine detail being brought out - magically almost! - rather than anything being added/manipulated.... unless that's the issue? (resolving detail that perhaps wasn't quite there)?   

I think it's a fine line, if the optics can't resolve detail, should it be there at all.

Link to comment
Share on other sites

26 minutes ago, tooth_dr said:

I have found with my own data, that I can push the sliders certainly above 10.  I wonder does it depend on the imaging scale and resolution of the camera, and quality of the data to begin with.  It doesnt look to me like it is adding anything, certainly nothing more than any of the other PS processes do.

I find that using a high pass filter with blend mode set to soft light has a similar although not quite so effective result. 

Link to comment
Share on other sites

13 minutes ago, Allinthehead said:

This is with sharpening slider set to 0

Looks to me that the de-noise is doing all the critical work. As you said above....it is a fine line! Nothing more (of my point of view) that I can add for now....I will watch with interest as I am sure there will be many more opinions to come.

Link to comment
Share on other sites

3 minutes ago, Kinch said:

Looks to me that the de-noise is doing all the critical work. As you said above....it is a fine line! Nothing more (of my point of view) that I can add for now....I will watch with interest as I am sure there will be many more opinions to come.

This is without any denoise and the sharpening set at 50

horse_test-4-lpc-cbg-Stmad2star222.png.14e99388004471e043d818d52277b4a0.png

Link to comment
Share on other sites

41 minutes ago, Kinch said:

I will never get the detail with a 5" refractor that someone with a 17" Planewave will get....so I know there are limits.  Overall it is that challenge that keeps me going.

This is a good point and it has led me to clarify my thoughts somewhat. I would never try to 'create' detail in an image that would or could never be there in the first place. As you say, gear has fundamental limits and detail beyond that is simply not achievable. 

1 hour ago, Allinthehead said:

I've no interest in using a tool like this if it's adding something that isn't in the data.

I suppose my stance is similar to Richard's above. By definition, de-noising is only destructive and that is fine to me. In theory, one would never require de-noising should they have enough data with a theoretical perfect camera. How de-noise is applied, whether that be via AI or regularly, is semantics. This is of course assuming that the AI does not create some structure in something that isn't there, by selectively processing certain parts of the image based on what it thinks should be there. I suspect that there is some aspect of this in the de-noise AI software, but it is hard to know.

It is a fine line and one that would require detailed analysis of the workings of the program to hash out, which can't / won't happen.

40 minutes ago, Allinthehead said:

This is with sharpening slider set to 0

Thanks for that, I've put together this comparison:  

image.png.c298e52496b316b069d4e9d47e771047.png

To my eye, both the de-noise and sharpening look like they could be adding 'detail' that isn't present in the original.

I suppose that at the end of the day, it is a tool in the astrophotographers toolbox that must be applied with full understanding that any undisclosed, AI-based processes may well destroy the integrity of the data, when compared to a more traditional method of noise reduction or sharpening such as a wavelet based approach. 

An interesting test would be to apply the tool to a stack of say, 10 subs, and to compare the result with an unmodified stack of 100 subs. That would be a good method in quantifying the destructive property of the tool.

EDIT: Updated image with sharpen only.

Edited by Spongey
  • Thanks 1
Link to comment
Share on other sites

37 minutes ago, Allinthehead said:

I think it's a fine line, if the optics can't resolve detail, should it be there at all.

It is a fine line.  We're all trying to extract as much as we legitimately can from our images and fundamentally, great images with loads of signal (like this one) make various stages of this easier, with better results. 

I think comparisons here are with an almost raw stack and a fully noise reduced + sharpened image?  Can "normal" processing resolve the same or close to the same detail?  I think you've now answered that above though 🙂

Edited by geeklee
  • Like 1
Link to comment
Share on other sites

Another way to evaluate how 'accurate' the tool may be in potentially 'creating' detail is to compare the image with one taken with a larger aperture scope. I've taken this image for reference and copied below for comparison (shot with a 304mm scope). I appreciate that this is not the best comparison as the reference image is narrowband, and looks like it has been through a fair share of noise reduction itself.

image.png.81428bc93a34b750ec03a1c6a1a5e6e0.png

 

Edited by Spongey
  • Like 1
Link to comment
Share on other sites

31 minutes ago, Spongey said:

Another way to evaluate how 'accurate' the tool may be in potentially 'creating' detail is to compare the image with one taken with a larger aperture scope. I've taken this image for reference and copied below for comparison (shot with a 304mm scope). I appreciate that this is not the best comparison as the reference image is narrowband, and looks like it has been through a fair share of noise reduction itself.

image.png.81428bc93a34b750ec03a1c6a1a5e6e0.png

 

What if they've also used denoise in that image🤣

  • Haha 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.