26 replies to this topic

[unimatrix0]

Having just finished my M31 of 38 and half hours and running 3 different versions of it and not sure if I should be celebrating or book it away as a learning experience, 

and just to know that "this is what I get if I do this", because I'm not totally happy about it. 

 

I also got 3 other DSO's yet to process, but those are emission nebulae so I'll see what I got. 

 

 

I came to some conclusions though. 

 

I also looked around on astrobin and I am not seeing a huge variation between the images with similar number of hours and similar equipment. 

 

Actually the images look quite the same, with some people pulling either harder on the saturation curve or the sharpening curve, but in overall the images look very similar. 

 

I actually found images that are much shorter (6-8hrs) that look better*, but I think I might have to credit the processing skill of the photographer and possibly the darker sky available, not the number of hours spent. 

 

I also understand, that I may have set my expectations too high.  

 

I have to say the only reason maybe worth imaging for very long  would be to have a greater number of subs that allows a much harsher quality control. 

 

There are more subs to be excluded without the fear of not having enough. 

 

Also, considering that I have imaged M31 many times before and my 5-10 hours of previous attempts aren't much far behind in detail and color. 

Also consider, that I used a triband filter, which is a limiting factor compared to a UV/IR.
BUT, also consider that I have imaged M31 for over 20 hours before with a UV/IR fitler only and I got a very similar result. 

 

With very bright DSOs' I think the "point of diminishing returns" sets in much earlier, then let's say a faint nebula that will require at least 15 hours to start out with. 

 

What I also think, if someone does image many DSO's with many more hours than the average imager, perhaps a wider field of view to reveal dust clouds that are invisible under 20 hours or Ha and OIII regions that don't really pop out unless we sit on the target for a while. 

 

For now, loaded my subs up again into the subframe selector process in Pixinsight and culling the subs heavy handed and see what I get if I re-stack, how it's gonna compare to what I got.

 

So, what does anyone think?  Do  you have an M31 image with 35+ hours worth of subs and what does it look like? 

Has anyone imaged something with many hours, but weren't blown away by the results? 

 

astrobin image, click for larger : 

 

Edited by unimatrix0, 11 November 2024 - 01:10 PM.

[Kolenka]

At some point you will wind up in the realm of diminishing returns when it comes to SNR. When that is depends on a lot of factors: brightness of the object, brightness of the sky, etc. So it's hard to do direct comparisons without knowing the conditions. I need a lot more integration time at home than at a star party for the same equipment, and bright objects need less integration time than dim ones. 

I don't have the link off hand, but there have been folks that have put together calculators to estimate the actual SNR you can expect to get with certain conditions, equipment and integration times. 

[lattitrail]

Not my image but I was browsing astrobin last night and found this 200+ hour image:https://app.astrobin...search?i=mxzpp9

 

Interstellar space is actually full of hydrogen at something like one atom per cubic centimeter. With enough exposure time, eventually one will capture it. Is it worth the time to achieve a reddish background glow? I guess that's left to individual opinion. In this case they were going for the O3 arc, so perhaps that makes it worth it here.

[Zambiadarkskies]

Not my image but I was browsing astrobin last night and found this 200+ hour image:https://app.astrobin...search?i=mxzpp9

 

Interstellar space is actually full of hydrogen at something like one atom per cubic centimeter. With enough exposure time, eventually one will capture it. Is it worth the time to achieve a reddish background glow? I guess that's left to individual opinion. In this case they were going for the O3 arc, so perhaps that makes it worth it here.

This is probably going to sound weird, but those crazy length Andromeda images with the Ha background do nothing for me as my brain says that there could be a galaxy out there but it would be on a dark background... Ironically I have never done any visual beyond binoculars.  The galaxy could be as detailed as anything, and even show nebula.  But the minute the background is red my mind tells me that it looks wrong.  To a certain extent this is also why I am not the world's biggest fan of SHO imaging.  

 

To go to Frank's question, I at times see a lot of images as exercises in technical brilliance that do not grab me.  It is a bit like the old adage in photography that the amount of effort and craft that you sink into an image does not automatically make it a 'good' image.  The person viewing is not looking through that lens of how much effort was put in.  The same applies with me when it comes to some of the insane astro landscape stuff that people do - think Barnard's loop popping over the dolomites.  A straight milky way shot I can get - tracked and blended.  But add in Ha regions and I am lost.  It is just not what I would see. Or even imagine I could see.  

 

This is not me saying that SHO is more or less real than anything else.  Or saying that 200Hr images of M31 are less real.  It is simply what my brain tells me when I see an image and what appeals to me.  

 

Again, going to Frank's question, for me the issue is that there are a lot of  amazing Andromeda or Orion pictures with incredible colour that has been teased out with interference filters over dozens if not hundreds of hours.  Some blow my mind and some don't.  Familiarity breeds contempt.  I would rather see some obscure object that I am not familiar with.  

 

Orion is another one - where people say "amazing, you have tamed the core".  And to me it looks like a hot mess in lurid pink from 15 years ago when HDR was all the rage.   

 

A great image Frank - I like it.  And yes, I have done big integrations and been left less that satisfied.  The Eagle nebula was one this year.   

[mehresman2]

More [quality] integration time = better SNR. The way I understand it, is is a fact, not an opinion.

 

Whether that increase is worth it to you, the artist, is a personal choice. You do you!

[Kolenka]

This is probably going to sound weird, but those crazy length Andromeda images with the Ha background do nothing for me as my brain says that there could be a galaxy out there but it would be on a dark background... Ironically I have never done any visual beyond binoculars.  The galaxy could be as detailed as anything, and even show nebula.  But the minute the background is red my mind tells me that it looks wrong.  To a certain extent this is also why I am not the world's biggest fan of SHO imaging.  

 

To go to Frank's question, I at times see a lot of images as exercises in technical brilliance that do not grab me.  It is a bit like the old adage in photography that the amount of effort and craft that you sink into an image does not automatically make it a 'good' image.  The person viewing is not looking through that lens of how much effort was put in.  The same applies with me when it comes to some of the insane astro landscape stuff that people do - think Barnard's loop popping over the dolomites.  A straight milky way shot I can get - tracked and blended.  But add in Ha regions and I am lost.  It is just not what I would see. Or even imagine I could see.  

 

On one level, I get that. There's something to be said for capturing things in a way that more closely matches what our eyeballs let us experience. Beauty is in the eye of the beholder, after all. 

 

On another I wind up feeling the opposite. The finding of the OIII arc for example, tells us something we didn't know about the universe before. So from the perspective of discovery and understanding, these exercises in technical brilliance grab me because it reveals something I've either not known before, or showing me something I've not seen before. While it's one thing to know about the medium that surrounds galaxies, it's another to actually witness it. And it's also interesting to be able to tease out the composition of nebulae and learn from it. 

 

I don't think there is a wrong way to approach it. 

[bbasiaga]

I have an image of Veil, I took two nights.  About 6 good hours each night.  I can't seem to get the 12hr integration to look as good as the 6hr night.  The subs look pretty similar.  The HFR is about the same.  The forecast for seeing was about the same....I don't know what's up.

 

The other thing is, once you get to 10 hours you have to go 4x to get double the SNR.   That's 40 hours.   Diminishing returns on the extra hours.   I have an image of the Jellyfish, where you can see some more details in the fainter areas after doubling the exposure time (40% SNR increase, give or take).  But the main nebula looks about the same.  I think the big differences lie there - in the really faint things hard to pull out of the background the extra time is more impactfull.  This is not every target.  

[SteveL42]

Your M31 IS gorgeous - don't be unhappy with it!  I did a similar one last month to see what I could get, about 30 hours total:

 

 

Taken in Bortle 5 skies.  But an 8 hour version would probably look 90% as "good".  

 

The sky darkness/quality will be a HUGE factor in what you get vs imaging time, somewhere there's a chart that shows the needed imaging time for equivalent SNR for different Bortle classes.  It's about 10:1 (IIRC) from my home to our clubs dark sky site (Bortle 2).  So, 3 hours there would have given me similar SNR in theory.

[archiebald]

With the scarcity of cloudless / moonless nights this year, I tend to shoot the minimum integration time to get something that can be viewed acceptably on a 27 inch monitor at 1 meter viewing distance.  The most I've ever tried was a three night run totaling 23 hours on the Elephant Trunk.

[Zambiadarkskies]

I have an image of Veil, I took two nights. About 6 good hours each night. I can't seem to get the 12hr integration to look as good as the 6hr night. The subs look pretty similar. The HFR is about the same. The forecast for seeing was about the same....I don't know what's up.

The other thing is, once you get to 10 hours you have to go 4x to get double the SNR. That's 40 hours. Diminishing returns on the extra hours. I have an image of the Jellyfish, where you can see some more details in the fainter areas after doubling the exposure time (40% SNR increase, give or take). But the main nebula looks about the same. I think the big differences lie there - in the really faint things hard to pull out of the background the extra time is more impactfull. This is not every target.

I have had that so often where I process an image and then add a couple more nights of integration and then utterly fail (even with the extra integration time) to improve on the quick process.

[psienide]

This is about 30 hours. I did notice a difference as things progressed that made it seem worth it to me. I'm imaging in heavy light pollution, so I need more time to bring out the faint stuff. But even then I think my LP drowns out the faintest edges.

I've also compared with images even only 6 hours total that look way better and made me wonder if it's really worth the effort. Either way, it's effort I have no choice but to be proud of!

 

 

Edited by psienide, 11 November 2024 - 11:23 PM.

[gsaramet]

My money is firmly on the SNR. 

 

However, keep in mind I shoot from B8-9. Basically I can't overshoot

 

I do have a project that's over 100h already - it's the Garlic, you can see it in my gallery. I have two projects running now - one will probably go 100h - the Spaghetti, the other well over - the Bat and Squid. Because that stuff is *faint*. 

 

Now, if your slice of sky has some shiny object, it's up to you (and your sky) what you want to capture and show. Say it's M42. Is it the Trapezium area? Is it the great nebula? Is it the integrated flux nebula? If it's M31, it begins again: is it the LRGB? Do you want to paint some Ha on it? Do you want to show the Oiii ring? Who knows, maybe there is a Sii tetrahedron that shows only at 10k hours. Or a He Klein bottle which shows only over 100k hours

 

Study case: my Spaghetti (not extremely faint). At about 10h I did an integration, stretched HARD and got some ugly noise bands in Oiii. Posted the WIP, checked other images, realized it was not noise, but signal.

 

Bottom line: longer integration = better SNR. The advantage in SNR becomes evident only in the faint areas. Do *you* want those areas on your image or not? 

[Distant]

That's a great image and certainly should not be seen as disappointment, unless your bar is extremely high.

I am actually working on a long-hours (60+) M31 project myself at present.
Usually I would aim at something around 12-16 hours but for M31 I just wanted to have a 'really nice and special' version without too many compromises - and along the way also see how much of a difference such an integration-heavy project really makes.
But admittedly, I have exactly the same concern you are experiencing now: realizing after having put in so much effort that one could have probably gotten the same with much less integration time.

 

Like you I had browsed Astrobin prior to starting on M31 just to get a feeling for how I should plan it out, how many hours per filters etc.
My reference is this one because the author is using somewhat similar equipment https://www.astrobin...bbe/?q=m31 oiii

Of course there are many variables, so comparison is extremely difficult - especially if you factor in post processing skills which probably has the biggest influence. While I am not expecting anything like the reference as result, I really want to try to visualize some of the OIII.

My current project status looks like this:

Subs taken is already after removing bad frames.
The Ha/OIII are shot with 3nm narrowband filters at my Bortle 8 home, while the broadband are taken from a Bortle 4 dark site.

I do not have much experience yet so it is a huge leap of faith regarding invested time, and I am already bracing for some disappointments. In that regards, I'd be happy to get advice on above plan and whether there is any potential to reduce some without any penalty.

Coming back to the OP, sorry there is not much for me to contribute yet with my project still ongoing and no image to share. Just felt like commenting you are not alone - and will definitely come back to check this thread as I am curious on other peoples' opinions.

[ShiftSix]

All I have to say is I commend your effort.  My A.D.D. starts kicking in around the 1 hour mark and begging me to see what else I can point the scope at lol 

[unimatrix0]

I have had that so often where I process an image and then add a couple more nights of integration and then utterly fail (even with the extra integration time) to improve on the quick process.

I have a couple of those.  

Last year with the Iris Nebula I just couldn't put my finger on what was wrong. Then I removed about 10 hours of subs that I made later and the image was better. I have no idea what exactly was causing it. I blamed the flats, I blamed seeing, I went through a whole bunch of scenarios I can think of, but the additional subs just would just destroy the image.  I stacked it like 4 times with different options and re-shot the flats and biases, and then at one point I quit on it and left it the way it was, like 10 hours instead of 20. 

Edited by unimatrix0, 12 November 2024 - 08:20 AM.

[unimatrix0]

All I have to say is I commend your effort.  My A.D.D. starts kicking in around the 1 hour mark and begging me to see what else I can point the scope at lol 

Yep, that was me too a few years ago, I did 1hr -2hr images of everything.   I still do that time to time, but also because of my limited view of the sky (2 trees on East and West has grown extremely tall in the past few years) I do 2-3hrs a night  and then I have to switch targets.  

I do 2-3hrs of that and switch again. 

 

Essentially I image 2 or 3 targets a night, each a few hours and repeat it as many times as the weather permitting. As I have figured out, it can get messy, because if the flats fail somehow and I don't have the correct ones, that night may have to go to the dumpster. 

 

Still my best pictures so far- that I personally like are no more than 1 or 2 nights and be done with it, instead of few hours each night and  stretching over weeks. 

Edited by unimatrix0, 12 November 2024 - 08:33 AM.

[ShiftSix]

Yep, that was me too a few years ago, I did 1hr -2hr images of everything.   I still do that time to time, but also because of my limited view of the sky (2 trees on East and West has grown extremely tall in the past few years) I do 2-3hrs a night  and then I have to switch targets.  

I do 2-3hrs of that and switch again. 

 

Essentially I image 2 or 3 targets a night, each a few hours and repeat it as many times as the weather permitting. As I have figured out, it can get messy, because if the flats fail somehow and I don't have the correct ones, that night may have to go to the dumpster. 

 

Still my best pictures so far- that I personally like are no more than 1 or 2 nights and be done with it, instead of few hours each night and  stretching over weeks. 

 

If you had to guess, what do you think that sweet spot is in terms of total hours on target & image quality? 

 

I'm having a ton of fun with the "under an hour challenge" but I definitely can see where data is missing in my images.   

Edited by ShiftSix, 12 November 2024 - 08:45 AM.

[Zambiadarkskies]

If you had to guess, what do you think that sweet spot is in terms of total hours on target & image quality? 

 

I'm having a ton of fun with the "under an hour challenge" but I definitely can see where data is missing in my images.   

I try and always aim for 10 hours plus.  Big bright ones like M31 and Carina can be less.  But double digits is generally my goal.  

[unimatrix0]

If you had to guess, what do you think that sweet spot is in terms of total hours on target & image quality? 

 

I'm having a ton of fun with the "under an hour challenge" but I definitely can see where data is missing in my images.   

Don't get me wrong- many of my images are around 8-15hrs I aim to get at least 10+ and if I can do 30+ that's fantastic as I said- just for the reason that I can make the sorting/selecting of subs even harsher, depending on eccentricity star shape and noise etc. 

 

. But I also have a bigger failure rate in that category- meaning - there is a good chance that my multi-night image possibly not going to process nicely. 

 

Whether it's the seeing or some other failure with something, my straight shot (one night after another with no cloudy days between) with my fast Newtonian is a sure bet to process out easier and with less headaches than my slower 80mm triplet, which need a lot more time, or waiting days or weeks for a clear sky and having 3-4 days of clouds between the sessions.

 

 

I mean look at what it take to make this Andromeda picture.  This is not something I like to do, but more like this is what I can do. 

 

I just can't have more imaging time per night on the galaxy then a few hours. 



 

Attached Thumbnails

• 
• 

Edited by unimatrix0, 12 November 2024 - 10:17 AM.

[unimatrix0]

Now, look at this Andromeda

 

https://astrob.in/t2c5m2/B/

 

Note that it was done in less than 4 hours with my Newtonian. Same camera, same Antlia triband filter. 

 

Here is a screenshot for demonstration

 

 

 

Attached Thumbnails

• 
• 

Edited by unimatrix0, 12 November 2024 - 10:13 AM.

[psienide]

Don't get me wrong- many of my images are around 8-15hrs I aim to get at least 10+ and if I can do 30+ that's fantastic as I said- just for the reason that I can make the sorting/selecting of subs even harsher, depending on eccentricity star shape and noise etc. 

 

. But I also have a bigger failure rate in that category- meaning - there is a good chance that my multi-night image possibly not going to process nicely. 

 

Whether it's the seeing or some other failure with something, my straight shot (one night after another with no cloudy days between) with my fast Newtonian is a sure bet to process out easier and with less headaches than my slower 80mm triplet, which need a lot more time, or waiting days or weeks for a clear sky and having 3-4 days of clouds between the sessions.

 

 

I mean look at what it take to make this Andromeda picture.  This is not something I like to do, but more like this is what I can do. 

 

I just can't have more imaging time per night on the galaxy then a few hours. 



 

Don't feel bad. Mine was almost 2000 subs...
 

[Andros246]

Sweet spot is 6-8 hours IMO

Really comes down to your processing ability.

Edited by Andros246, 12 November 2024 - 12:13 PM.

[Jared]

Having just finished my M31 of 38 and half hours and running 3 different versions of it and not sure if I should be celebrating or book it away as a learning experience, 

and just to know that "this is what I get if I do this", because I'm not totally happy about it. 

 

I also got 3 other DSO's yet to process, but those are emission nebulae so I'll see what I got. 

 

 

I came to some conclusions though. 

 

I also looked around on astrobin and I am not seeing a huge variation between the images with similar number of hours and similar equipment. 

 

Actually the images look quite the same, with some people pulling either harder on the saturation curve or the sharpening curve, but in overall the images look very similar. 

 

I actually found images that are much shorter (6-8hrs) that look better*, but I think I might have to credit the processing skill of the photographer and possibly the darker sky available, not the number of hours spent. 

 

I also understand, that I may have set my expectations too high.  

 

I have to say the only reason maybe worth imaging for very long  would be to have a greater number of subs that allows a much harsher quality control. 

 

There are more subs to be excluded without the fear of not having enough. 

 

Also, considering that I have imaged M31 many times before and my 5-10 hours of previous attempts aren't much far behind in detail and color. 

Also consider, that I used a triband filter, which is a limiting factor compared to a UV/IR.
BUT, also consider that I have imaged M31 for over 20 hours before with a UV/IR fitler only and I got a very similar result. 

 

With very bright DSOs' I think the "point of diminishing returns" sets in much earlier, then let's say a faint nebula that will require at least 15 hours to start out with. 

 

What I also think, if someone does image many DSO's with many more hours than the average imager, perhaps a wider field of view to reveal dust clouds that are invisible under 20 hours or Ha and OIII regions that don't really pop out unless we sit on the target for a while. 

 

For now, loaded my subs up again into the subframe selector process in Pixinsight and culling the subs heavy handed and see what I get if I re-stack, how it's gonna compare to what I got.

 

So, what does anyone think?  Do  you have an M31 image with 35+ hours worth of subs and what does it look like? 

Has anyone imaged something with many hours, but weren't blown away by the results? 

 

astrobin image, click for larger : 

 

So, a few things to keep in mind...

• You already mentioned diminishing returns... The brighter the object (particularly the parts of the object that contain the details you care about), the fewer hours you need to get to a "reasonable" SNR. If I were going after NGC 869 and NGC 884 (The Double Cluster), for example, I would probably not need more than a couple hours of data with relatively short exposures to help me maintain star saturation, and even then I would probably throw out the lower quality half of my data in terms of FWHM. The same is true for most globular clusters. Diminishing returns happen really quickly with intrinsically bright objects, so you can quickly start shifting priorities to the quality of the individual subs.
• Many objects gain in detail and subtlety as you are able to reach deeper and deeper into the shadows. Sometimes this is true even for objects that are pretty bright. Perhaps the most famous example would be the galaxy pair of M81 and M82 (Bode's Galaxy and the Cigar Galaxy respectively). Pulling out the red "galactic winds" around M82 require a fair amount of exposure time as does the IFN that seems to surround both galaxies. So, even though these two objects rival M31 for surface brightness, they definitely benefit a lot from lots of integration time.
• Even M31 can benefit from a lot of integration time depending on what details you are interested in drawing out. If you want to be able to show how the galactic halos of M31 and M110 appear to be overlapping, there is no question you need a lot of exposure. If you just want to show the dust lanes in M31 with reasonable contrast? You might be able to get away with just a few hours of integration time. Depends on your skies and your aperture. 
• It is important to keep in mind light pollution can play a profound role in how much integration time you need for a given object. I live in Oakland, California under Bortle 8 skies (when there is no moon). My telescope is housed remotely under Bortle 3 skies. I could easily need 20x as much time imaging from home with broadband filters as I need at my remote site in order to get similar SNR. So, just knowing that someone spent 60 hours on a given target doesn't remotely tell the whole story. 
• Your processing skill matters also, particularly your ability to handle stretching while maintaining contrast.
• Recent improvements in software tools have allowed astrophotographers to "get away" with a lot less integration time recently. In particular, the AI tools that can reduce "blur" help a lot with contrast in fainter areas, and AI tools that address noise can let you stretch your data farther. So, where you might have needed thirty hours in the past to keep fainter sections of M31 from looking noisy after stretching, now you might be able to get a similar result with just six or eight hours.
• Since your particular example was M31, keep in mind that larger objects often require mosaics, depending on the scope and camera, and mosaics can artificially push the integration time very high very quickly. I have a twelve panel mosaic of the Horsehead region that isn't all that deep in any given panel even though the total integration time is over one hundred hours. Worth it? Honestly, it was the only way for me to get the field of view I wanted with reasonable detail.

My personal approach--and everyone is different--is to spend at least one full night on any given object. After that, I do a "quick and dirty" integration and post processing to see if I have managed to capture what I wanted, or whether I need more data to draw out the details I am looking for. These days, more often than not I find I do want a lot more data. I'm not certain what my average is, but it's certainly in excess of thirty hours per target. But, I have the luxury of a remote observatory that can collect data on clear nights without a lot of attention from me, so it isn't much of a sacrifice to decide, "Yeah, another couple nights should do the trick!" If I had to drive somewhere, setup and babysit my scope, then drive back from each observing session, I suspect I would make very different decisions.

[TXLS99]

I have a couple of those.  

Last year with the Iris Nebula I just couldn't put my finger on what was wrong. Then I removed about 10 hours of subs that I made later and the image was better. I have no idea what exactly was causing it. 

I dont know what processing software you are using, but I also noticed an improvement in my images when I started getting more aggressive culling my subs rather than just stacking everything.

 

This video helped a lot  https://www.youtube....h?v=mLmtVcgaQZo

 

 

Even if you dont use Pixinsight the approach is the same.  Removing subs with lower star counts helped me a lot, its like basically removing "dirty" subs from the final stack

[AstroVagabond]

I did a first for me during October and November new moon periods. I collected, under Bortle 3/4 skies, 25 hours plus of the Spaghetti Nebula and 25 hours plus of the Cygnus Loop. Up until then I generally stopped around 12 hours.

 

For me my time investment was worth it and I even wished I had gone to 30 - 35 plus hours on each image.

 

https://www.astrobin.com/8xsdea/

 

https://www.astrobin.com/rr5ffh/

 

This is a have it your way hobby, it's our wallets funding it, and each of gets to call the shots. In my view there is no controlling authority.  I'm now firmly in the more time on target camp shooting for qaulity per image versus quantity of images.

 

~ Bill