Bubble Nebula Comparison

[Rodd]

10 hours ago, Kinch said:

Hi Rodd.....I think you will take one thing from this thread.....most people do not "Pixel Peep" and it is just first impression that gets the likes....as someone said, it is just a simple 'push of a button'. Like you, I can spend hours on an image because one thing or another just annoys me about it, but the end result generally wont make it any more of a winner for those that just flick through the images. I have seen it time & time again, to get 'likes' you need colourful, attention grabbing images - not presented in highest resolution. To get satisfaction though, you correct the 'nasties' that you see on your high resolution monitor....so that you feel satisfied with the processing. The important bottom line is...don't substitute 'Likes' for satisfaction.....or perhaps putting it another way.....don't seek to get satisfaction from getting 'likes'.....get it from your creation.

It seems that the top image here is more appealing to most but when you open the door to the higher resolution its bad points become very apparent. The 2nd image is an improvement but someone also pointed out where that can been improved. So neither are perfect (and in truth a lot of APODs are not either) but I can tell that the second sits better with you. So you have improved on a previous effort.....take satisfaction out of that. 🙂 

Thanks Kinch....yes the second sits better with me...but I am definitely not satisfied with it!  If I ever was to create the perfect image, I think it would be akin to the little dog that finally caught the school bus...what then?  What you say is true--but there are a group of very experienced processors on this forum whose comments always include going beyond the screen image (clicking on the image at least once--if the image is good, and people like it, I think most "imagers" look at full resolution--especially those acclaimed members whose comments  are most coveted.  For me, looking at full resolution is prerequisite for calling an image "good"  

Rodd

[Filroden]

Hi Rodd

This is a tough one. I like the presentation of the first - both the wider context and the bolder saturation of the colours. However, I think the second is the better of the two images. The main things that swing it for me are you have colourless stars in the first, whereas the second image has coloured stars. They just make the second image look more natural. The other thing that swung it is that the first image appears "blotchy" in the background - almost like one "frequency" of structure has been over processed. The very largest scale structure looks good, and the medium and smallest scale structures are exquisite - but the large structures look "wrong". I don't see that issue in the second (but to be fair, the issue is most pronounced in the wider background which you've cropped out).

[Kinch]

9 minutes ago, Rodd said:

For me, looking at full resolution is prerequisite for calling an image "good"  

Indeed - I agree 100%. But if it is 'Likes' that you are after - it does not have to be THAT good.

 

11 minutes ago, Rodd said:

yes the second sits better with me...but I am definitely not satisfied with it

You are not satisfied with it because at Hi Res you see the problems. So do that group that you mention below....

 

15 minutes ago, Rodd said:

a group of very experienced processors on this forum whose comments always include going beyond the screen image

 

Never mind the 'Likes' - aim to get comments from those experienced processors. If they comment...it means they are interested and having a closer look. 🙂 

[Rodd]

5 hours ago, Freddie said:

It strikes me that the problem with your images is that however many likes you get you will always want more and therefore you will never be happy with your image. Over the months you have made many comments about the likes you get (or don’t get) and threatened to leave the forum at one stage as people weren’t giving you the approval you were looking for through your images. Maybe  its time to determine what number of likes would satisfy your need for praise from others and once you have made changes to the image to achieve that number you can move on to your next target. Hope these thoughts help you come to terms with this as the number of likes appears to be something you spend a lot of time worrying about.

Not so......The issue is seeing an image get 27 likes and 25 comments and then posting an image that I feel is as good or better--and when asked people tend to agree that it is very good--that gets 6 likes and a comment or two.  And its not praise from  others that I am after--it is improvement.  Likes verify that an image is thought decent by those who have more experience and skill than I--and comments about what is wrong with an image allow me to improve.  To be clear--I associate likes with Astrobin (comments are much more rare there, at least for me).  But someone years ago convinced me that likes on SGL were good because after many comments--what more can be said about an image other than "looks great mate"  Hence clicking the like button is easier.   And, just so you don't think I make empty threats...I did leave the forum for a couple of months (torture for me, as I am a verifiable astrophotohaulic...and for those who may regard such jests an affront to a serious condition that they feel I am making light of...I jest not.  I returned and things were much better.  Now, I accept the fact that few comment or hit like (in most cases), so I post comparisons like this when I have doubt.  That does not mean I still do not wonder why this is.  And to be clear--I stated the "like issue" in this thread to explain why I made the post.  After that, my discussion of likes has been in comment to other folks comments.  Besides, what it to you if I like likes?  I thought that what forums like this are for (for those looking to improve).  Otherwise, posting an image is more bragging than looking for ways to improve.  People will say  CC desired, or please do not hesitate to provide CC.  Well, for me, that goes without saying

Rodd

[Rodd]

5 hours ago, Laurin Dave said:

Hi Rodd

I like them both a lot but with a slight preference for the bottom one, it looks more natural (whatever that means) to me..  Perhaps combine the two sets of data for an even better one.  

Dave

Now there is an Idea--unfortunately I only have the JPEG of the first image after my hard drive crashed.......but fear not, I am already planning my next attempt (oh god!)

Rodd

[Kinch]

3 minutes ago, Rodd said:

Likes verify that an image is thought decent by those who have more experience and skill than I

Disagree...……for me, when someone is motivated to hit the like button, it just means that on first sighting the image....it pleases them. It says nothing about whether or not they have more experience or (processing) skill than you.

[Rodd]

4 hours ago, Gina said:

I haven't looked at the full images but I like the top image better even if it has been processed almost to death.  I don't know which was taken with which camera and I would like to know but I think the differences are more in the processing than the camera.  At least that's the impression I get.  Both are superb images.

I too think you place too much emphasis on Likes - the important thing is what you like best.  Do it for yourself rather than others.  OTOH I know we all like to get lots of Likes!

Thanks Gina--but its not likes per se that I am after.  It is confirmation from those I respect as imagers that I am on the right track.  Its all about creating as good an image as possible.  Above all else I am looking for CC, advise, and yes, support.  Likes are a great way for acknowledging a "good job".  I have only been doing this a short time, and likes from the experts really are nice.   Every time I look at one of my images I feel differently--so my judgement is a slippery thing.

Rodd

[Rodd]

4 hours ago, Rob Sellent said:

@Rodd I apologise if I don't visit the imaging section that much - do imagers check out the sketchers section etc? - but I just wanted to say I find your photography outstanding and a joy to see. The two images supplied in this thread are absolutely gorgeous

Thanks Rob.  To be honest, I do not visit the sketching section very often (and I am a drawer!).  The problem is visual observing in my sky always leaves me wanting--there are only a few targets that can really be seen.  After you look at them a dozen times....what then?  Things would be different from Bortle 1 or 2 I am sure.  

Rodd

[Rodd]

4 hours ago, wesdon1 said:

Absolutely stunning image! Amazing work sir!

Thanks wesdon1!

Rodd

[Rodd]

4 hours ago, vlaiv said:

Hint for everyone wishing to know camera used for each image:

Both images were taken at same focal length but different pixel scale (one is more "close in" than the other, and other has better SNR for same integration time - "resolution at aperture" in works), none was binned.

Cameras in question have different pixel size ....

A complicating factor is I cropped one--and.....I actually software binned the ASI 1600 then resampled to be almost normal size (100 pixels short I believe--I think anyway--I processed the image so many times I know I did that with many--and I think with teh final version).  Thanks, by the way--I have made this technique part of my work flow when appropriate!.  Anyway Vlad, you know which image was taken with each camera....I will spill the beans to the others but I want to let everyone know so will quote as many as I can.

Rodd

[Rodd]

2 hours ago, glowingturnip said:

for me I'd say the first, but only presented in that kind of image-scale, it definitely starts suffering at full res.  I like the detail in the clouds round the bubble that don't usually get brought out much, and the dark 'hollowness' of the bubble itself.

The second is the superior shot on close examination, but for me it is definitely too soft.  Did you deconvolute ?

I did--but not the first.  Its a tough pixel scale for a 5" scope in bortle 5/6.  

Rodd

[Rodd]

2 hours ago, vlaiv said:

There seems to be prevalent trend in comments here if I'm not mistaken?

People love first image at screen size because it provides more depth and sharpness. People also love second image on 1:1 inspection.

Ideally you want the combination of the two in your final rendition (not images but things that people prefer). Second image feels a bit soft because it is oversampled, at 1000mm, 3.8um pixel size will give resolution of 0.78"/px.

At 5.4um pixel size and 1000mm focal length - you will be sampling at 1.11"/px - this is closer to optimum sampling rate for your conditions (that I suspect will be order of 1.2-1.3"/px with 130mm aperture and good mount under average seeing in NB).

Btw, SNR ratio between these two resolutions is about x1.42, so things that are very low SNR in one image let's say SNR of 3.5 (barely distinguished from the noise) will have SNR of 5 (perfectly acceptable to be included in the image as legitimate signal that will be fairly smooth with some denoising).

@Rodd

If you happen to have all the data from the first image and all the data from the second image (meaning lights and calibration frames or at least calibrated lights for each channel) - maybe we could make "master image" by combining all the data at the same resolution - that being 1.11"/px?

We could do a little "collaboration" perhaps? I have fractional binning algorithm that works on calibrated subs rather than on final result (I think that it will be better that way), that actually does not produce "binned" subs, but splits subs in such way that makes more of them at target resolution.

We could either make all subs be sampled at 1.11"/px if you have all the data or maybe target 1.2-1.3"/px if you have the data for second image only, and see if you can process such stack to have features of both images - sharpness and depth at screen size and detail and sharpness (as opposed to softness) at 1:1 zoom?

 

Great Ideas Vlad--alas, all data from first image is gone.  The simple answer moving forward is to throw the reducer on the scope--which will yield a 1.12 arcsec/pixel resolution.  BUT--then I lose the scale--I suppose I could crop (or upsample)--but would the cropped/upsampled image look as sharp?  Or would I be right back here.   But I agree completely--.78 is too much for a 5" scope (regardless of its pedigree).  

Your software development is intriguing.  One problem I see is getting you 300 subs!   Does it work on stacks?

Rodd

[Rodd]

39 minutes ago, Filroden said:

Hi Rodd

This is a tough one. I like the presentation of the first - both the wider context and the bolder saturation of the colours. However, I think the second is the better of the two images. The main things that swing it for me are you have colourless stars in the first, whereas the second image has coloured stars. They just make the second image look more natural. The other thing that swung it is that the first image appears "blotchy" in the background - almost like one "frequency" of structure has been over processed. The very largest scale structure looks good, and the medium and smallest scale structures are exquisite - but the large structures look "wrong". I don't see that issue in the second (but to be fair, the issue is most pronounced in the wider background which you've cropped out).

Interesting Ken--thank you.  The reason I cropped the second image is becuase there was nothing there!  Just black space.  Thats because the sensitivity was not high enough to capture the outer, dim regions.  The image looked a bit odd with a wide dark boundary--so I cropped.

Rodd

[Rodd]

30 minutes ago, Kinch said:

Disagree...……for me, when someone is motivated to hit the like button, it just means that on first sighting the image....it pleases them. It says nothing about whether or not they have more experience or (processing) skill than you.

But the forum tells you who likes the image.  So when you get a like from one of the standard bearers, its a good indication that you are on the right track

Rodd

[Rodd]

ALL--The forum does not allow me to quote from multiple pages, so I could not ensure everyone who commented gets teh message--I hope you do.......The first image was with the STT-8300 and the second image was with the ASI 1600.  As Vlad indicated, the resolution of the second image is just too much for my sky/aperture.  I have found it works very well for the brightest objects--hence the bubble itself looks OK--but the surrounding nebula is a bit weak.  3.8um pixels are just too small for 1,000mm FL.  REDUCER TIME

 

Thanks to all that provided comments--all are appreciated.  Keep up the good work everyone....I find what we do astounding.  30 years ago it would have been beyond belief.  

Rodd

[MartinFransson]

First - the images: I like the first one best because of the field of view with the surrounding area. I does however feel a bit mottled in the darker parts when inspected at full res. The second image is smoother, but softer due to over sampling.

Second - the likes and comments: I´ve thought quite a lot about this myself since I´ve noticed the same thing. I think part of the problem is not being able to see thumbnails of the images. You have to click on a topic to even get a hint of the quality of the image. That means people (at least myself) probably look at subjects and go for something they want to see. If I´ve seen 10 images of M31, it is not likely that I´ll click another one. Next, you go by topic starter. If I see a new topic by someone who always posts mindblowing images, I´ll click. There comes the part about being in the "elite group" So I think it´s a combination. Regarding comments, a lot of the time I browse on my phone. Writing comments is a pain in the... but a like is easy to click away. However, if the topic starter has specifically asked a question seeking advice I am more willing to write something. And also, if I see that the topic starter is someone who already knows what they are doing I am less likely to write something.

[Rodd]

16 minutes ago, MartinFransson said:

First - the images: I like the first one best because of the field of view with the surrounding area. I does however feel a bit mottled in the darker parts when inspected at full res. The second image is smoother, but softer due to over sampling.

Second - the likes and comments: I´ve thought quite a lot about this myself since I´ve noticed the same thing. I think part of the problem is not being able to see thumbnails of the images. You have to click on a topic to even get a hint of the quality of the image. That means people (at least myself) probably look at subjects and go for something they want to see. If I´ve seen 10 images of M31, it is not likely that I´ll click another one. Next, you go by topic starter. If I see a new topic by someone who always posts mindblowing images, I´ll click. There comes the part about being in the "elite group" So I think it´s a combination. Regarding comments, a lot of the time I browse on my phone. Writing comments is a pain in the... but a like is easy to click away. However, if the topic starter has specifically asked a question seeking advice I am more willing to write something. And also, if I see that the topic starter is someone who already knows what they are doing I am less likely to write something.

All fair points.  The first thing I do is look for someone I know (by name)--and that is a lot of people by now.  Those are the threads I most often visit.  Doesn't really change my point.  As far as oversampled--I thought undersampling would yield softness--over sampling really does not take anything away--it just samples at what is available.  That's why its commonly said that oversamplig is better than undersampling becuase you don't lose anything.  So now I am confused.

Rodd

[MartinFransson]

1 minute ago, Rodd said:

All fair points.  The first thing I do is look for someone I know (by name)--and that is a lot of people by now.  Those are the threads I most often visit.  Doesn't really change my point.  As far as oversampled--I thought undersampling would yield softness--over sampling really does not take anything away--it just samples at what is available.  That's why its commonly said that oversamplig is better than undersampling becuase you don't lose anything.  So now I am confused.

Rodd

Don´t be confused, it might be that I mixed up the terminology. I have a bad case of man-cold right now. Fever does not exactly make me think more clearly

[vlaiv]

1 hour ago, Rodd said:

Great Ideas Vlad--alas, all data from first image is gone.  The simple answer moving forward is to throw the reducer on the scope--which will yield a 1.12 arcsec/pixel resolution.  BUT--then I lose the scale--I suppose I could crop (or upsample)--but would the cropped/upsampled image look as sharp?  Or would I be right back here.   But I agree completely--.78 is too much for a 5" scope (regardless of its pedigree).  

Your software development is intriguing.  One problem I see is getting you 300 subs!   Does it work on stacks?

Rodd

You are quite right about sharing 300 subs - even with faster internet connection than mine, sharing almost 20GB of data is not an easy thing to do (it would take me days to upload results on net connection that I currently have).

That is going to be real pain to transfer. We could do things the other way around if it is not too much involved for you. I could write a ImageJ plugin and explain step by step how to apply it to your subs. Transfer time in this case would be much less (few kb if that much since I would send you source code and ImageJ would compile it when installing the plugin), but it would require you to follow some number of steps to apply plugin to your subs. We can do that if you are willing.

It would however still produce smaller scale than you would like, unless you resize/upsample the image. This however won't bring any new detail in the image - just make things larger and a bit blurry.

This might be the reason why there is slight softness to the image - besides no detail being captured at this scale - once you upsample image softness starts to show at 1:1 even if baseline image (one that was upsampled) is properly sampled.

8 minutes ago, Rodd said:

All fair points.  The first thing I do is look for someone I know (by name)--and that is a lot of people by now.  Those are the threads I most often visit.  Doesn't really change my point.  As far as oversampled--I thought undersampling would yield softness--over sampling really does not take anything away--it just samples at what is available.  That's why its commonly said that oversamplig is better than undersampling becuase you don't lose anything.  So now I am confused.

Rodd

Undersampling does not yield softness - it yields lack of detail that could be possibly captured, but image should look sharp if viewed 1:1 without any sort of upscaling. It will look soft if you upsample it / enlarge things.

Oversampling on the other hand leads to softness when you look at it 1:1. Indeed it does no fail to capture any additional detail so no data is lost when oversampling, but same thing happens as when you properly sample image and upscale / enlarge - it becomes blurry because lack of the finest detail that could be displayed by single pixels (or rather local contrast).

This is why I advocate proper sampling - image should look sharp enough when viewed 1:1 (and also when viewed at screen size). Undersampled image will look similarly sharp at 1:1 (or screen size), but because it is so small scale - things won't be as resolved as with proper sampling. With oversampling - you get softness because things get really large in scale but with finest detail missing (not because of sampling but because you could not capture that finest detail due to other things - like seeing / aperture / guiding errors and associated blur).

Does this sense?

[astro mick]

Well Rodd,your certainly getting the comments at the mo.

I usually get none or one or two,and a few likes.

But I,m happy.

Mick.

[Rodd]

43 minutes ago, vlaiv said:

You are quite right about sharing 300 subs - even with faster internet connection than mine, sharing almost 20GB of data is not an easy thing to do (it would take me days to upload results on net connection that I currently have).

That is going to be real pain to transfer. We could do things the other way around if it is not too much involved for you. I could write a ImageJ plugin and explain step by step how to apply it to your subs. Transfer time in this case would be much less (few kb if that much since I would send you source code and ImageJ would compile it when installing the plugin), but it would require you to follow some number of steps to apply plugin to your subs. We can do that if you are willing.

It would however still produce smaller scale than you would like, unless you resize/upsample the image. This however won't bring any new detail in the image - just make things larger and a bit blurry.

This might be the reason why there is slight softness to the image - besides no detail being captured at this scale - once you upsample image softness starts to show at 1:1 even if baseline image (one that was upsampled) is properly sampled.

Undersampling does not yield softness - it yields lack of detail that could be possibly captured, but image should look sharp if viewed 1:1 without any sort of upscaling. It will look soft if you upsample it / enlarge things.

Oversampling on the other hand leads to softness when you look at it 1:1. Indeed it does no fail to capture any additional detail so no data is lost when oversampling, but same thing happens as when you properly sample image and upscale / enlarge - it becomes blurry because lack of the finest detail that could be displayed by single pixels (or rather local contrast).

This is why I advocate proper sampling - image should look sharp enough when viewed 1:1 (and also when viewed at screen size). Undersampled image will look similarly sharp at 1:1 (or screen size), but because it is so small scale - things won't be as resolved as with proper sampling. With oversampling - you get softness because things get really large in scale but with finest detail missing (not because of sampling but because you could not capture that finest detail due to other things - like seeing / aperture / guiding errors and associated blur).

Does this sense?

You cant work with the stacks?  That would be the easiest.  I am willing to try anything--lets plan on trying--but I need  to set aside some time and to mentally prepare.  Right now I am hurriedly trying to finish a couple of images then want to switch to the FSQ 106 and ASI 1600 because I have not used it since it was recollimated by TNR.  Its been a couple months and I really need to do it in case collimation wasn't the issue.  If I wait too long, they will not believe me--even though it has not been removed from the box since I got it back.  I am really tempted to try the TOA with teh reducer first, though.  But I know if I do that it will be months before I am willing to make another change.  Regarding oversampling--I really think its target dependent.  Targets with a high signal do not look quite as soft.  For example, here is a crop of my Wall image--.78 arcsec/pixel--cropped.  I am quite pleased with the detail in this image from a 5" scope.  Full separation in the double is not commonly achieved, even with much larger scopes.  Some targets are just easier it seems.  Some targets give me mottled backgrounds and high chroma noise--and some don't.  Maybe the seeing was better for this image.  I had a post about this Wall image being compared to the crop from the FSQ--didn't get much response...nothing like this thread.  But IMO, the TOA wall trounces the FSQ Wall though the FSQ Wall is brighter--it is over processed.  

Edited September 25, 2019 by Rodd

[Rodd]

17 minutes ago, astro mick said:

Well Rodd,your certainly getting the comments at the mo.

I usually get none or one or two,and a few likes.

But I,m happy.

Mick.

I am glad you are happy.  Have a work flow?

Roidd

[vlaiv]

13 minutes ago, Rodd said:

You cant work with the stacks?  That would be the easiest.  I am willing to try anything--lets plan on trying--but I need  to set aside some time and to mentally prepare.  Right now I am hurriedly trying to finish a couple of images then want to switch to the FSQ 106 and ASI 1600 because I have not used it since it was recollimated by TNR.  Its been a couple months and I really need to do it in case collimation wasn't the issue.  If I wait too long, they will not believe me--even though it has not been removed from the box since I got it back.  I am really tempted to try the TOA with teh reducer first, though.  But I know if I do that it will be months before I am willing to make another change.  Regarding oversampling--I really think its target dependent.  Targets with a high signal do not look quite as soft.  For example, here is a crop of my Wall image--.78 arcsec/pixel--cropped.  I am quite pleased with the detail in this image from a 5" scope.  Full separation in the double is not commonly achieved, even with much larger scopes.  Some targets are just easier it seems.  Some targets give me mottled backgrounds and high chroma noise--and some don't.  Maybe the seeing was better for this image.  I had a post about this Wall image being compared to the crop from the FSQ--didn't get much response...nothing like this thread.  But IMO, the TOA wall trounces the FSQ Wall though the FSQ Wall is brighter--it is over processed.  

In principle it can work with stack, but you will need to "restack" it, and results will probably be poorer than doing it with subs.

Idea is to split image into multiple subs (thus avoiding pixel blur that comes from larger pixels and preserving some of the sharpness in that stage). If you split subs, registration and stacking will do better job of optimizing everything then doing split on already stacked image. So yes, it will work on stack, but as I see it - better results would be had from doing it on each individual sub and then stacking those.

I don't have plugin written yet, so it will be couple of days before I have it - no rush in this, we can take it slowly step by step, once I finish plugin, I'll write a "tutorial" on how to apply it, and then you can do it when you have the time for it ...

Yes, there is no perfect sampling rate for all occasions - it really depends on night to night basis (or even on sub to sub basis) as seeing / guiding performance changes. Choice of target also plays a part both in brightness, but also in local contrast of the target - blur is really loss of local contrast, so not all targets will respond equally to the same amount of blur. Also position of the target in the sky relates to seeing effects as well - lower the target, more atmosphere you have between the target and the scope and more chance for seeing effects.

[Rodd]

3 hours ago, vlaiv said:

In principle it can work with stack, but you will need to "restack" it, and results will probably be poorer than doing it with subs.

Idea is to split image into multiple subs (thus avoiding pixel blur that comes from larger pixels and preserving some of the sharpness in that stage). If you split subs, registration and stacking will do better job of optimizing everything then doing split on already stacked image. So yes, it will work on stack, but as I see it - better results would be had from doing it on each individual sub and then stacking those.

I don't have plugin written yet, so it will be couple of days before I have it - no rush in this, we can take it slowly step by step, once I finish plugin, I'll write a "tutorial" on how to apply it, and then you can do it when you have the time for it ...

Yes, there is no perfect sampling rate for all occasions - it really depends on night to night basis (or even on sub to sub basis) as seeing / guiding performance changes. Choice of target also plays a part both in brightness, but also in local contrast of the target - blur is really loss of local contrast, so not all targets will respond equally to the same amount of blur. Also position of the target in the sky relates to seeing effects as well - lower the target, more atmosphere you have between the target and the scope and more chance for seeing effects.

Well Vlad--I took everything that was said to heart and tried a reprocess.  No--I did not bn 2x2.  I plan on trying that next.  My opinion is the bubble and nearby nebula came out better--but the surrounding space is too dark--such a fine line.  Well--on second thought not so sure--my screen is set toward the dark end.  What do you think?

Rodd

 

Crop-sharpened a bit more

 

Edited September 25, 2019 by Rodd

[Adam J]

16 hours ago, Rodd said:

Well Vlad--I took everything that was said to heart and tried a reprocess.  No--I did not bn 2x2.  I plan on trying that next.  My opinion is the bubble and nearby nebula came out better--but the surrounding space is too dark--such a fine line.  Well--on second thought not so sure--my screen is set toward the dark end.  What do you think?

Rodd

 

Crop-sharpened a bit more

 

That improved the second one but I think the blacks are looking a little blue at least on my monitor.