Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Galaxy Imaging Tips

astrophotography imaging
  • Please log in to reply
18 replies to this topic

#1 Chaz007

Chaz007

    Explorer 1

  • *****
  • topic starter
  • Posts: 94
  • Joined: 22 Sep 2016
  • Loc: Chicago Suburbs/Tucson AZ

Posted 17 February 2020 - 01:57 AM

For the past couple of months I’ve been working on my narrowband game, but as galaxy season approaches I’m on the verge of buying an RGB filter set (astronomik deep sky colors+l2.) It seems like there are a lot of key differences between narrowband and RGB galaxy imaging and I have a few questions before we’re in the thick of galaxy season...

 

So right now I’m typically getting about 5 hours of data per filter which is working out pretty well for me... However, I’ve seen quite a few galaxy images where there’s maybe an hour for r, g, and b, and a bunch of time for Ha and luminance. As far as percentage of integration time, how much time do you guys recommend spending on each filter. I also imagine Ha isn’t too fruitful for things like Markarian’s chain, but great for stuff like the cigar galaxy...

 

I’m also curious about binning. I bin for nebula imaging which resulted in a major improvement in SNR, image quality, and sheer signal. However, galaxies are pretty small so I’m not sure its worth sacrificing the detail.(I do drizzle though) I’ve also seen people bin on the color data and not on the luminance data. (Or maybe the other way around? I can’t remember)The other factor here though is that since I won’t be using narrowband filters I’ll be getting much more light to the sensor.

 

My final question is about LP protection. I held on to my Idas lps d2 after selling my dslr just in case, and I’ve considered putting it in front of my filter wheel. Right now I’m thinking I won’t use it because I’ve heard that can really decrease the amount of data you can get, but without it I’ll be completely unprotected, especially for luminance. I suppose I can experiment, but what do you guys think? If it helps I DO have some great gradient removal software for photoshop(astroflat pro) but I imagine that can only do so much...

 

Any help/additional tips would be greatly appreciated!

 

Oh, and I’m running a William optics Z73 with ASI 183mm.


Edited by Chaz007, 17 February 2020 - 02:11 AM.

  • Swordfishy likes this

#2 Astrodymium

Astrodymium

    Explorer 1

  • *****
  • Posts: 83
  • Joined: 04 Sep 2019

Posted 17 February 2020 - 02:11 AM

Even with a mild LP filter like the IDAS LPS D2, Optolong L Pro, Orion SkyGlow, etc, you are blocking about 50% of the visible spectrum of light. You'd have to be getting a massive increase in contrast to make up for losing 50% of the signal from a broadband target.

 

I wouldn't use it on galaxies.


Edited by Astrodymium, 17 February 2020 - 02:11 AM.

  • pfile and Swordfishy like this

#3 happylimpet

happylimpet

    Skylab

  • *****
  • Posts: 4,450
  • Joined: 29 Sep 2013
  • Loc: Southampton, UK

Posted 17 February 2020 - 04:07 AM

So right now I’m typically getting about 5 hours of data per filter which is working out pretty well for me... However, I’ve seen quite a few galaxy images where there’s maybe an hour for r, g, and b, and a bunch of time for Ha and luminance. As far as percentage of integration time, how much time do you guys recommend spending on each filter. I also imagine Ha isn’t too fruitful for things like Markarian’s chain, but great for stuff like the cigar galaxy...

 

I’m also curious about binning. I bin for nebula imaging which resulted in a major improvement in SNR, image quality, and sheer signal. However, galaxies are pretty small so I’m not sure its worth sacrificing the detail.(I do drizzle though) I’ve also seen people bin on the color data and not on the luminance data. (Or maybe the other way around? I can’t remember)The other factor here though is that since I won’t be using narrowband filters I’ll be getting much more light to the sensor.

 

My final question is about LP protection. I held on to my Idas lps d2 after selling my dslr just in case, and I’ve considered putting it in front of my filter wheel. Right now I’m thinking I won’t use it because I’ve heard that can really decrease the amount of data you can get, but without it I’ll be completely unprotected, especially for luminance. I suppose I can experiment, but what do you guys think? If it helps I DO have some great gradient removal software for photoshop(astroflat pro) but I imagine that can only do so much...

 

 

General consensus is that LP filters, in this era of LED lights, dont help.

 

Dont bin then drizzle, that makes no sense. Dont sacrifice resolution; thats the main goal.

 

Put some hours in. I use a trick of imaging in Luminosity and with a 610 longpass (essentially infrared) as my B and R, and interpolating for G. This works brilliantly for getting rudimentary colour on super faint targets as you're pulling in so much light, but I guess with a refractor you dont have this option of imaging in the IR - maybe I'm wrong?

 

Beware of redshift, which moves h-alpha out of the bandpass of narrowband filters at quite a modest distance. M82, M101, M31 etc are all fine though.



#4 Stelios

Stelios

    Cosmos

  • *****
  • Moderators
  • Posts: 8,443
  • Joined: 04 Oct 2003
  • Loc: West Hills, CA

Posted 17 February 2020 - 04:11 AM

When I only spend an hour or so for R/G/B, I get color noise that's hard to remove which shows up as speckling even after the LRGB combine. 

 

I think what works for me is to get at least 90 min per filter, and combine the RGB with the Lum into a super-luminance (then process the super-luminance for detail, and recombine with the RGB). 

 

I always take some Ha, but not too much, as I only use it to enhance Red, and color noise gets diluted out. 

 

I don't bin with CMOS cameras. I dislike LP filters. 



#5 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 11,950
  • Joined: 14 Jan 2013

Posted 17 February 2020 - 08:41 AM

Aside from the question of whether LP filters even help with galaxies, they often introduce color shifts. For example, beautiful yellows tend to end up looking school bus yellow (orange).



#6 schmeah

schmeah

    Fly Me to the Moon

  • *****
  • Posts: 5,476
  • Joined: 26 Jul 2005
  • Loc: Morristown, NJ

Posted 17 February 2020 - 09:01 AM

I've been imaging galaxies from heavy light pollution for years.

 

1) I tried an IDAS-LPS but found that my results were better without it.

2) I try to capture at least as much luminance (typically 3-4 hours) or more than total RGB (45-60 mins per channel). I can typically complete a galaxy image in one evening.

3) I bin my RGB (detail is really all in the luminance IMO) but I have a CCD. Leave the luminance un binned unless seeing is poor or your image scale requires it (mine is 0.65" without FR) 

4) Exposure time Gradient removal techniques are key in LP. Get Gradient Xterminator if you don't have it already.

 

Derek


  • milkychris and OhmEye like this

#7 milkychris

milkychris

    Lift Off

  • -----
  • Posts: 24
  • Joined: 09 Jun 2019
  • Loc: St Petersburg, FL

Posted 17 February 2020 - 06:42 PM

I've been imaging galaxies from heavy light pollution for years.

 

1) I tried an IDAS-LPS but found that my results were better without it.

2) I try to capture at least as much luminance (typically 3-4 hours) or more than total RGB (45-60 mins per channel). I can typically complete a galaxy image in one evening.

3) I bin my RGB (detail is really all in the luminance IMO) but I have a CCD. Leave the luminance un binned unless seeing is poor or your image scale requires it (mine is 0.65" without FR) 

4) Exposure time Gradient removal techniques are key in LP. Get Gradient Xterminator if you don't have it already.

 

Derek

Your images are very inspiring, especially if they come from heavy light pollution


Edited by milkychris, 17 February 2020 - 06:44 PM.

  • Swordfishy likes this

#8 Stephen Kennedy

Stephen Kennedy

    Surveyor 1

  • *****
  • Posts: 1,933
  • Joined: 03 Jul 2014
  • Loc: California

Posted 17 February 2020 - 11:30 PM

Why are you using a 73 mm refractor to image galaxies when you have a 5" F/8 reflector? Except for a couple, galaxies are small and faint objects.  You are going to want to use as much aperture and focal length as you can to image galaxies like those in the Virgo Cluster.



#9 Chaz007

Chaz007

    Explorer 1

  • *****
  • topic starter
  • Posts: 94
  • Joined: 22 Sep 2016
  • Loc: Chicago Suburbs/Tucson AZ

Posted 17 February 2020 - 11:49 PM

I can’t tell you how much I agree, but unfortunately it’s a pretty cheap reflector (spherical mirror with bird jones lens to correct) and I don’t think my mount could handle it. In the future my plan is to get a massive triplet refractor and throw a reducer on it for nebulae. That’d also require a mount upgrade... Now that you’ve got me thinking about it though it may not hurt to give it a try. smile.gif

 

For the time being I’m going to focus on targets such as the Leo triplet, M81/82, Markarian’s chain, and a few galaxies that are surprisingly large. That should easily keep me occupied for this spring at the rate I’m imaging.

 

Thanks for all the help! I think I have a pretty good idea of what I’m going to do now. One last question though: In general, what kinds of total integration time are you guys getting on galaxies? I’m thinking I’ll probably be at about ten hours on average, maybe a bit less, with half of that being lum. That should let me finish a project in a weekend, though I can of course stretch that if I’m finding it isn’t enough.



#10 OhmEye

OhmEye

    Messenger

  • -----
  • Posts: 455
  • Joined: 15 Sep 2019
  • Loc: Western NY Southern Tier

Posted 18 February 2020 - 08:35 AM

2) I try to capture at least as much luminance (typically 3-4 hours) or more than total RGB (45-60 mins per channel). I can typically complete a galaxy image in one evening.

3) I bin my RGB (detail is really all in the luminance IMO) but I have a CCD. Leave the luminance un binned unless seeing is poor or your image scale requires it (mine is 0.65" without FR)

I happened to use almost exactly the same approach last week from my bortle 3-4 site. My viewing angle with trees mostly limits me to 3-4 hours max per target and I did 4 hours on M81/M82, NGC 4565, and almost 2 hours on M13 that night. My 4 hour breakdown was about 45m per RGB and the rest on L. My 1x1 pixel scale was 0.67, I did all the L at 1x1 and all the RGB at 2x2. I have CMOS, so I don't get the slight read noise advantage of binning CCD, but 2x2 trades RGB resolution I probably don't need for double the SNR with the idea that the detail comes from the L channel. I haven't done final processing but I'm fairly pleased with the data.

 

I'm just preparing for my first galaxy season and have decided to learn what I can with my refractor this season before choosing a dedicated galaxy scope for the 2021 season. I also want to try the method of combining LRGB to make a super luminance channel before adding back in the RGB, ideally to directly compare both methods with the same data.

 

My pixel scale with the FR is 0.67", I can swap the FR for a FF for galaxy season to get 0.53" and if seeing is less than good I can bin 2x2 for 1.06" which seems reasonable to me on a F7 930mm.


  • schmeah and Swordfishy like this

#11 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 11,950
  • Joined: 14 Jan 2013

Posted 18 February 2020 - 10:49 AM

I happened to use almost exactly the same approach last week from my bortle 3-4 site. My viewing angle with trees mostly limits me to 3-4 hours max per target and I did 4 hours on M81/M82, NGC 4565, and almost 2 hours on M13 that night. My 4 hour breakdown was about 45m per RGB and the rest on L. My 1x1 pixel scale was 0.67, I did all the L at 1x1 and all the RGB at 2x2. I have CMOS, so I don't get the slight read noise advantage of binning CCD, but 2x2 trades RGB resolution I probably don't need for double the SNR with the idea that the detail comes from the L channel. I haven't done final processing but I'm fairly pleased with the data.

 

I'm just preparing for my first galaxy season and have decided to learn what I can with my refractor this season before choosing a dedicated galaxy scope for the 2021 season. I also want to try the method of combining LRGB to make a super luminance channel before adding back in the RGB, ideally to directly compare both methods with the same data.

 

My pixel scale with the FR is 0.67", I can swap the FR for a FF for galaxy season to get 0.53" and if seeing is less than good I can bin 2x2 for 1.06" which seems reasonable to me on a F7 930mm.

Since you are using CMOS, there is no gain in Signal/Noise when binning on camera compared to when binning after the fact (you mention you don't get the read noise advantage so I think you are aware of this). There is of course a gain in space required for the image. But you are giving up resolution by doing so. This matters if you ever want to do something like using the RGB to create synthetic L or to mix with the L.


Edited by Madratter, 18 February 2020 - 01:27 PM.

  • Swordfishy likes this

#12 OhmEye

OhmEye

    Messenger

  • -----
  • Posts: 455
  • Joined: 15 Sep 2019
  • Loc: Western NY Southern Tier

Posted 18 February 2020 - 11:36 AM

Since you are using CMOS, there is no gain in Signal/Noise when binning on camera and when binning after the fact (you mention you don't get the read noise advantage so I think you are aware of this). There is of course a gain in space required for the image. But you are giving up resolution by doing so. This matters if you ever want to do something like using the RGB to create synthetic L or to mix with the L.

My understanding of binning CMOS comes from here: https://forums.sharp...topic.php?t=262

 

Binning CMOS does improve Signal/Noise. Binning trades spacial resolution for improved SNR. The gains are better for CCD, but it does still have value for CMOS when used appropriately. That's my understanding and experience so far, but I'd like to understand where you disagree with Robin Glover. In his words: "CMOS style software binning increases the SNR in proportion to the N (of NxN binning), so 2x2 doubles the SNR, 3x3 triples it, etc"

 

Of course, wanting to use the RGB for something other than just color information, like creating any channel for the purpose of contributing to detail, would not be an appropriate use of binning. I think the use for only contributing color to strong clean unbinned luminance is an appropriate use, especially when the image scales involved are well matched to the seeing conditions. The other night, my binned RGB was 1.34"/pixel which was actually probably close to the seeing, and my unbinned L was 0.67"/pixel which was slightly oversampled. If I'm understanding everything correctly, I think it actually worked out about as ideal as it can for this use case.



#13 Swordfishy

Swordfishy

    Ranger 4

  • *****
  • Posts: 340
  • Joined: 08 Feb 2019
  • Loc: Miami, FL

Posted 18 February 2020 - 11:53 AM

Since you are using CMOS, there is no gain in Signal/Noise when binning on camera and when binning after the fact (you mention you don't get the read noise advantage so I think you are aware of this). There is of course a gain in space required for the image. But you are giving up resolution by doing so. This matters if you ever want to do something like using the RGB to create synthetic L or to mix with the L.

So, the concensus is, don't bother binning for RGB since you won't benefit at all?

 

 I have read through this whole thing since even at my image scale 1:1 (1.06) 400FL I can do decently on most galaxies due to the super tiny pixels of my ASI183MM.

 

 I was thinking 3-5 hours luminance and 1 hour each RGB. 

 

 I am in the muck... Bortle 9++ and LED assault, but I am confident I can still pull something worth looking at ;-)

 

 I feel bummed when it is 10pm and there is NOTHING for me to image in my very limited NE slice of "heaven". Now I feel better about it with Galaxies. I feel renewed with hope until we can once more start playing with DSOs.



#14 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 11,950
  • Joined: 14 Jan 2013

Posted 18 February 2020 - 12:07 PM

My understanding of binning CMOS comes from here: https://forums.sharp...topic.php?t=262

 

Binning CMOS does improve Signal/Noise. Binning trades spacial resolution for improved SNR. The gains are better for CCD, but it does still have value for CMOS when used appropriately. That's my understanding and experience so far, but I'd like to understand where you disagree with Robin Glover. In his words: "CMOS style software binning increases the SNR in proportion to the N (of NxN binning), so 2x2 doubles the SNR, 3x3 triples it, etc"

 

Of course, wanting to use the RGB for something other than just color information, like creating any channel for the purpose of contributing to detail, would not be an appropriate use of binning. I think the use for only contributing color to strong clean unbinned luminance is an appropriate use, especially when the image scales involved are well matched to the seeing conditions. The other night, my binned RGB was 1.34"/pixel which was actually probably close to the seeing, and my unbinned L was 0.67"/pixel which was slightly oversampled. If I'm understanding everything correctly, I think it actually worked out about as ideal as it can for this use case.

 

 

Where you are misunderstanding is this. CMOS software binning absolutely increases the SNR as stated by Robin. But whether that software binning occurs on camera or off camera doesn't matter. Downloading it 1x1 and then rebinning 2x2 in software will double the SNR.

 

Here is an experiment you can do for yourself. Take a random unstretched sub from your camera. Clone it, and then resample to 50% size in both axis. Then run the noise evaluation script in PixInsight.

 

You will find the Noise is about halved in the software binned version. It won't be exact since the Noise Evaluation Script is estimating the noise and it doesn't always distinguish noise properly from signal. But it will be close.

 

Here I have done this with a random sub of mine.

 

SoftwareBin2x2 20200218.JPG

 

And here is the noise evaluation. In this particular random sample it thinks it did better than 2 for 1 but again that is because of errors in distinquishing signal from noise in the script.

 

NoiseEvaluation 20200218.PNG



#15 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 11,950
  • Joined: 14 Jan 2013

Posted 18 February 2020 - 12:10 PM

So, the concensus is, don't bother binning for RGB since you won't benefit at all?

 

 I have read through this whole thing since even at my image scale 1:1 (1.06) 400FL I can do decently on most galaxies due to the super tiny pixels of my ASI183MM.

 

 I was thinking 3-5 hours luminance and 1 hour each RGB. 

 

 I am in the muck... Bortle 9++ and LED assault, but I am confident I can still pull something worth looking at ;-)

 

 I feel bummed when it is 10pm and there is NOTHING for me to image in my very limited NE slice of "heaven". Now I feel better about it with Galaxies. I feel renewed with hope until we can once more start playing with DSOs.

For CMOS there is no signal to noise benefit for doing it on camera as opposed to after the fact. Both are using software binning. That said, there is the advantage of smaller files to work with, and somewhat reduced processing time working with those smaller files.


  • Swordfishy likes this

#16 Swordfishy

Swordfishy

    Ranger 4

  • *****
  • Posts: 340
  • Joined: 08 Feb 2019
  • Loc: Miami, FL

Posted 18 February 2020 - 12:19 PM

For CMOS there is no signal to noise benefit for doing it on camera as opposed to after the fact. Both are using software binning. That said, there is the advantage of smaller files to work with, and somewhat reduced processing time working with those smaller files.

Great! Thanks for clarifying. Here is what I got after about 8 hours of Luminance. Extremely cropped obviously.

 

 

 

 Cool, so I will shoot LRGB at 1x1.

Attached Thumbnails

  • 49498297937_c090acee0f_c.jpg

  • milkychris likes this

#17 OhmEye

OhmEye

    Messenger

  • -----
  • Posts: 455
  • Joined: 15 Sep 2019
  • Loc: Western NY Southern Tier

Posted 18 February 2020 - 01:14 PM

Where you are misunderstanding is this. CMOS software binning absolutely increases the SNR as stated by Robin. But whether that software binning occurs on camera or off camera doesn't matter. Downloading it 1x1 and then rebinning 2x2 in software will double the SNR.

I'm not misunderstanding that, I know it doesn't matter when the software binning happens, it can be in the camera driver or later in processing. If I had a reason to want both 1x1 and 2x2 for the same subs of course I would image at 1x1 and bin after the fact. That's actually my intent for comparing the same data for 2x2 RGB vs. 1x1 RGB contributing to a super-luminosity channel.

 

I misunderstood when you said "there is no gain in Signal/Noise when binning on camera and when binning after the fact" when what you meant was there is no SNR difference *between* on camera and off camera. Now I understand you agree there *is* gain in SNR when binning, when your wording implied otherwise. Actually, there is an important difference between binning on camera and off camera for many CMOS cameras, but I didn't think it relevant to the original conversation. "on camera" for CMOS means hardware binning, and software binning is after the fact after downloading, either by the driver or by later processing. For many cameras, the "on camera" binning uses fewer ADC bits, which is almost never desirable. For example, "on camera" binning for the ASI1600 and ASI183 uses 10bit ADC instead of 12bit, so I would never use "on camera" binning for my cameras.

 

Thanks for clarifying the misunderstanding, I think I'm on the same page now. waytogo.gif



#18 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 11,950
  • Joined: 14 Jan 2013

Posted 18 February 2020 - 01:14 PM

By the way, it may possibly have occurred to some of you that there is another alternative to binning. It is what I commonly do with my RGB data. Specifically, you can convolve it. This also combines the data from multiple pixels but in a rather more complicated way.

 

This is also subject to experimentation. I did this again with the same starting sub as above. Here I did a rather mild convolution as such things go.

 

Convolution 20200218.JPG

 

Here is the signal to noise estimate from PixInsight:

 

Convolution Noise 20200218.PNG

 

Note that this made a much bigger impact than the factor of 2 SNR improvement you get by doing 2x2 software binning. That is because convolution with these parameters is actually using far more pixels to mix together.

 


  • Swordfishy likes this

#19 Madratter

Madratter

    Voyager 1

  • *****
  • Posts: 11,950
  • Joined: 14 Jan 2013

Posted 18 February 2020 - 01:28 PM

I'm not misunderstanding that, I know it doesn't matter when the software binning happens, it can be in the camera driver or later in processing. If I had a reason to want both 1x1 and 2x2 for the same subs of course I would image at 1x1 and bin after the fact. That's actually my intent for comparing the same data for 2x2 RGB vs. 1x1 RGB contributing to a super-luminosity channel.

 

I misunderstood when you said "there is no gain in Signal/Noise when binning on camera and when binning after the fact" when what you meant was there is no SNR difference *between* on camera and off camera. Now I understand you agree there *is* gain in SNR when binning, when your wording implied otherwise. Actually, there is an important difference between binning on camera and off camera for many CMOS cameras, but I didn't think it relevant to the original conversation. "on camera" for CMOS means hardware binning, and software binning is after the fact after downloading, either by the driver or by later processing. For many cameras, the "on camera" binning uses fewer ADC bits, which is almost never desirable. For example, "on camera" binning for the ASI1600 and ASI183 uses 10bit ADC instead of 12bit, so I would never use "on camera" binning for my cameras.

 

Thanks for clarifying the misunderstanding, I think I'm on the same page now. waytogo.gif

Original statement has been edited to be hopefully more clear. I think we are indeed on the same page.


  • Swordfishy likes this


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics





Also tagged with one or more of these keywords: astrophotography, imaging



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics