Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
I went to a dark sky site last night for the first time, and made some comparisons from my home site.
My home site sky is 4.25 - 4.50 mag, and the dark site is 5.25 - 5.50.
Each shot is 7x10 minutes, and conditions on both nights were excellent.
The first image is the stack from DSS with no processing.
The second shot has quick basic processing. I set the black point on each, then did the same curve stretches.
--------------------
Arran
My main website
My blog
|
gatsbyiv
member
Reged: 03/29/09
Posts: 67
Loc: Whitehouse Station, NJ
|
|
I find posts like this to be enormously educational as a novice. Nothing speaks more clearly than an example. (It also makes me feel better that I can't coax any detail out of M101 from my light polluted skies!)
Someone should compile all of these A/B examples that have been posted on CN (LP vs. non-LP skies, 80mm vs. 102mm shots, F7 vs. F10, etc...) into a book.
Thanks for posting this, Aza!
--------------------
-----------------------
Astro Telescopes 102ED
Astro Telescopes 80ED
Meade LXD75 Mount
Canon Rebel XSi
Just enough knowledge to be dangerous
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
I was shocked, as I had never been to darker skies before.
Previously I had 12 hours of M101 data, but this single hour of dark sky data is of about the same quality!
Being in the UK, I dont get many oppurtunites for imaging. So it seems to me, it is worth the hours drive to get to the darker skies.
--------------------
Arran
My main website
My blog
|
Patrick
Postmaster
Reged: 05/16/03
Posts: 7809
Loc: Franklin, Ohio
|
|
Hi Arran,
Thanks for posting the very informative images. I had the same experience this weekend imaging M81, once from the house and once from our clubs dark sky site. The dark sky site was so much better that it makes the 45 minute trip there seem trivial (except for the time and energy it takes to pack everything up). Maybe I should be focusing on ways to keep everything ready to go at a moments notice rather than on ways to cut through the lp in my neighborhood. 
Patrick
--------------------
10" f/6 Truss Tube Newtonian
Celestron C6S-GT SCT
AT66ED Refractor
Oberwerk Ultra 15x70 Binocular
475B Geared Tripod & 501HDV Head
Celestron Regal 8x42 Binocular
Canon XSi; Meade DSI;SPC900-NC
Vixen GP2 Photo Guider Mount
My Astronomy Pages
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Yeah I know what you mean Patrick!
As I dont have a permanent set up at home, I feel as if I should make the trip each time I image in future.
However being out on your own in the cold with nothing to do, is a bit different from sitting indoors and checking things every once in a while 
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Hi Arran,
Your images are interesting, but don't really tell the whole story. The fact is that since the images from the dark and not-so-dark sites have the same total exposure times, the amount of signal you have in the galaxy is the same for each. The difference is the amount of signal in the sky glow, but the galaxy signal sits on top of the sky glow, so it can be extracted. It takes more stretching, and the room left for stretching is different because of the higher baseline, but all of the details you captured from the dark site are there in the image from the not-so-dark site as well.
Dark skies are a distinct advantage, obviously, but not a reason to defer imaging until you can make the trip IMHO.
I edited your second JPEG (with GIMP because I'm currently on a Linux PC, and I'm not very familiar with GIMP) to demonstrate the fact that all of the details are there in both images. With some careful manipulation to optimize both (and starting with RAW or TIFF files), it would be difficult to tell the two apart.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
turbo399
sage
Reged: 12/21/07
Posts: 226
Loc: Michigan
|
|
Nice images. That is a very interesting comparison. I have just moved from a light polluted city to a country location. I would be all set if the weather was was an equal improvement.
Terry
--------------------
TMB 130SS F7 Refractor
TMB 80SS F6.3 Refractor
Celestron C14
10" F5 Astrograph/Newt
Stellarvue 80ED Nighthawk NGD
Vixen 80ED
Celestron CGE PRO Mount
QHY8
Canon 450D modified
Orion Autoguider
http://www.flickr.com/photos/terryhancock/
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Hi Don
You have shown that very well! For me though, it was so much easier processing the dark sky image. I could get the image looking good with a few stretches. To get the same detail showing in my LP image, I had to make a right mess of the rest of the image.
It seemed to me that the detail was much more forthcoming, without having to butcher the image. I would like to see what you could do with the full 12 hour stack I have!
To me the one hour image is on par with the 12 hour image I have.
--------------------
Arran
My main website
My blog
|
Patrick
Postmaster
Reged: 05/16/03
Posts: 7809
Loc: Franklin, Ohio
|
|
Don,
That's interesting too. Did you try to do the same processing on the dark sky version? Given what you've said, what are the advantages of dark skies?
Patrick
--------------------
10" f/6 Truss Tube Newtonian
Celestron C6S-GT SCT
AT66ED Refractor
Oberwerk Ultra 15x70 Binocular
475B Geared Tripod & 501HDV Head
Celestron Regal 8x42 Binocular
Canon XSi; Meade DSI;SPC900-NC
Vixen GP2 Photo Guider Mount
My Astronomy Pages
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Hi Don
You have shown that very well! For me though, it was so much easier processing the dark sky image. I could get the image looking good with a few stretches. To get the same detail showing in my LP image, I had to make a right mess of the rest of the image.
It seemed to me that the detail was much more forthcoming, without having to butcher the image. I would like to see what you could do with the full 12 hour stack I have!
To me the one hour image is on par with the 12 hour image I have.
Hi Arran,
You're exactly right, it is much easier to process images where the brightness of the sky glow is low in comparison to the subject. But the real advantage of dark sky sites comes with being able to increase the exposure times, thereby catching more photons, while keeping the sky glow under control. If you had shot the dark site images at 20 minutes instead of 10 minutes, the sky glow brightness might be about the same as in 10 minute exposures from the light polluted site, but you would have captured twice as many photons from the galaxy. Then you would be faced with more difficult post-processing to separate the subject from the sky glow, but the results would be very noticeably better.
I think it's a shame when beginners get the idea that they can't do AP from home because of the light pollution. In all but the worst conditions, that simply isn't true. And the skills you can gain in post-processing by dealing with lots of sky glow in your images will pay huge benefits when you ARE able to get to a dark site.
I suggest buying a book on astrophotography image processing, and if you use Photoshop (there's really nothing better for post-processing IMHO), perhaps buying some pre-made tools like Noel Carboni's Photoshop actions for AP. But if you have the patience and inclination, you can do everything Noel's tools do with nothing but Photoshop out of the box. Obviously Noel did
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
jmasin
sage
Reged: 12/22/08
Posts: 489
Loc: Murphy, TX (DFW)
|
|
Arran,
Thanks for your post. Undoubtedly pulling detail out of the soup is harder
I'm surprised you were able to get 10 minutes from your "LP" sky. Were you using a filter on the LP sky and no filter on the dark site? You don't mention any camera settings. I can say from my home site 2 minutes (ISO1600, so 4 minutes ISO800) nearly completely saturates the frame with LP. Any more than that is a near washout. I'd love to get 10 minutes out of my home site!!
Thinking about the impact of LP is straightforward if you think about a single pixel example.
At a LP site, say a pixel receives 1 DSO photon per minute, and 0.5 LP photons per minute. For a two minute exposure you have 2 DSO photons and 1 LP photon that you have to differentiate between.
At a dark site, say a pixel receives 1 DSO photon per minute, and 0.1 LP photons per minute. A ten minute exposure gives you 10 DSO photons and 1 LP photon. A much easier distinguished difference.
Like Don says, going to longer exposures is the real gain of dark site. Above example is way oversimplified because there may be some fainter details of the DSO that send the same photons/min as the LP. Those may not be distinguishable from the LP for any reasonable amount of integration time.
--------------------
Cheers,
Jon
--------------
Orion Atlas 8" (203mm f/4.9 on Atlas EQ-G)
Tasco 60mm (~35 years old and still kicking!)
Nikon D300, 80, 70 & many lenses
http://www.pbase.com/jmasin/astro
Edited by jmasin (04/27/09 01:24 PM)
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Don,
That's interesting too. Did you try to do the same processing on the dark sky version?
Hi Patrick,
The real advantage of dark skies is the ability to shoot longer subs. Longer subs always increase the ratio of signal to noise, but under light polluted skies not only do longer subs require more skill in post processing, but the dynamic range left over after the sky glow is clipped is smaller, so less stretching is possible before artifacts start to show up.
If there were a site where there were absolutely zero light pollution, you would still need to shoot multiple subs and combine them to reduce noise, but if your subs were long enough the stacked image would need no stretching.
Quote:
Given what you've said, what are the advantages of dark skies?
Patrick
No, I didn't try stretching the dark sky image. With these small JPEGS it would be futile. If Arran can post the stacked 16-bit images somewhere it would be interesting to see how much can be made of both of them.
I would expect somewhat better results from the dark sky site, but not dramatically better, since the total integration time is the same in both.
Cheers,
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
I'm surprised you were able to get 10 minutes from your "LP" sky. Were you using a filter on the LP sky and no filter on the dark site? You don't mention any camera settings. I can say from my home site 2 minutes (ISO1600, so 4 minutes ISO800) nearly completely saturates the frame with LP. Any more than that is a near washout. I'd love to get 10 minutes out of my home site!!
Me too! I would have to go down to ISO 200 to get decent 10 minute exposures from my home site. That would actually be better from the noise perspective than 5 minutes at ISO 400 or 2.5 minutes at ISO 800, due to the fact that longer exposures have higher SNR than shorter ones with the same exposure value (time x ISO). But the lower ISO with longer exposures would mean getting much fewer subs in one night, and potentially running into problems with tracking/guiding and vibrations.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Very interesting!
About the settings:
Both images were ISO 800.
I used a CLS filter for the LP shot, and none for the dark shot.
I have some experience in processing already, and use photoshop along with the actions and some plug ins.
However I was unable to make the 1 hour LP image anywhere as good as the 1 hour dark image. In fact it was onyl with 8 hours or so, that I could get the LP image as good as the 1 hour dark shot.
Here is my final processed image, which is 12 hours of LP shots plus the 1 hour dark shot.
I would love someone to have a process of my data to show me how much better it can be!
Where can I upload a 50mb file?
--------------------
Arran
My main website
My blog
|
Patrick
Postmaster
Reged: 05/16/03
Posts: 7809
Loc: Franklin, Ohio
|
|
Quote:
I would expect somewhat better results from the dark sky site, but not dramatically better, since the total integration time is the same in both.
I don't doubt what you're saying, but a lot of what I've read (and experienced) is that light polluted skies require magnitudes (pun intended) more integration time than dark sky sites. Samir Kharusi in particular states that as sky fog goes up, the integration times required to achieve the same SNR as the dark site would go up by the difference in the stellar magnitudes.
The problem with more integration time and shorter exposures (reduced for sky fog) is that a lot more subframes will be required. More subframes equates to a lot more post processing time. I've noticed a huge difference in my post processing time going from a 12mp camera from an 8mp camera, let alone adding a huge stack. I'll have to get a new computer to handle it. Of course, post processing by itself is not that big a deal...just get it started and go take a lunch break.
Don't get me wrong...I want to be able to image from my back yard, but I'm just not sure if it's worth the effort.
Patrick
--------------------
10" f/6 Truss Tube Newtonian
Celestron C6S-GT SCT
AT66ED Refractor
Oberwerk Ultra 15x70 Binocular
475B Geared Tripod & 501HDV Head
Celestron Regal 8x42 Binocular
Canon XSi; Meade DSI;SPC900-NC
Vixen GP2 Photo Guider Mount
My Astronomy Pages
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Just found somewhere to upload the tifs, will pass on the links shortly for anyone that wants to have a play.
--------------------
Arran
My main website
My blog
|
Mike Clemens
Post Laureate
Reged: 11/26/05
Posts: 4278
Loc: Wasilla, Alaska 61N
|
|
Simple, take 400 - 800 exposures !
<ducks>
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Hi Arran,
That is a really great image! You did a very good job of post-processing obviously, and I'm not sure I could make it any better.
Where I have gained the most experience is in processing images with substantial light pollution, and I would love to have a go at yours.
I think there are some free sites that you can upload files to so that others can download them. Perhaps someone else will read this and provide the info. If not, I may be able to make an FTP site available temporarily.
Cheers,
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Mike, it was seeing your M101 that really made me think something must be wrong for me! So I headed to the dark skies
--------------------
Arran
My main website
My blog
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Thanks Don
I am currently uploading my 2 comparison stacks, plus the full 13 hour stack.
I would be interested to see what people make of them!
--------------------
Arran
My main website
My blog
|
jmasin
sage
Reged: 12/22/08
Posts: 489
Loc: Murphy, TX (DFW)
|
|
I know I couldn't top that last one! That's a great photo.
--------------------
Cheers,
Jon
--------------
Orion Atlas 8" (203mm f/4.9 on Atlas EQ-G)
Tasco 60mm (~35 years old and still kicking!)
Nikon D300, 80, 70 & many lenses
http://www.pbase.com/jmasin/astro
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Hi Patrick,
I've read Samir's essays and he is obviously quite knowledgeable, but I don't agree with all of his conclusions.
Regarding sky glow, it is not noise, it is part of the signal. It contains noise, mostly shot noise, as do all parts of the image, but the SNR due to shot noise increases with illumination, so the overall SNR of a raw image, or a stack of raw images, from a light polluted site is actually higher (better) than the SNR of the same image or stack, of the same subject, acquired with the same gear and with the same exposure time and ISO at a darker site. But in either case you must remove the sky glow and then stretch what's left. With larger amounts of sky glow there will be less range remaining for the stretching process after you remove the sky glow. In a worst case, assuming a 12-bit RAW image, which has 4096 discrete units of signal, high levels of light pollution may reduce the available signal for the subject by half to 2048 units. This will make a noticeable difference in the ability to stretch the data, and the SNR of the final stretched image will suffer.
But 2048 counts of sky glow is very high - the image would appear nearly white when seen on the camera's LCD display, in Photoshop or in other software that uses a logarithmic gamma curve. I routinely push my subs until the sky glow hump is between the middle and about 2/3 of the way to the right, and I never have more than about 1000 counts out of 4096 in the sky glow.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
http://www.mediafire.com/?sharekey=2ffa80c7805fde6c391d7d881749d3a7e04e75f6e8ebb871
Here are the stacked tifs. I would love to see some results on the 13 hour stack. But also maybe some processed comparisons between the LP and dark stacks.
--------------------
Arran
My main website
My blog
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
Regarding exposure, So I have a Canon 40D, and someone mentioned in Canon's Digital Photo Pro in the tools menu, there's a linear checkbox, So I've been messing around checking my exposure in linear mode, and I've been shooting 6-10 minute subs at 1600 (~f/6), where in logarithmic mode on really clear nights the longest I could do was 3-4 minutes.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Changing DPP to linear mode doesn't change the data at all, it just changes the way the data is distributed within the display range. So the bottom line is you thought you couldn't go over 3-4 minutes before (due to the appearance of very bright sky glow), but you could really have gone 6 to 10 minutes. The trick then is to stretch the data to remove the sky glow and enhance the subject without losing anything of value. This may be easier if the data is distributed linearly, but again, the log distribution contains exactly the same data. This is the reason that you can switch back and forth between linear and log mode at will.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
Edited by DonR (04/27/09 04:07 PM)
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
Quote:
Changing DPP to linear mode doesn't change the data at all, it just changes the way the data is distributed within the display range. So the bottom line is you thought you couldn't go over 3-4 minutes before (due to the appearance of very bright sky glow), but you could really have gone 6 to 10 minutes. The trick then is to stretch the data to remove the sky glow and enhance the subject without losing anything of value. This may be easier if the data is distributed linearly, but again, the log distribution contains exactly the same data. This is the reason that you can switch back and forth between linear and log mode at will.
You are correct, I thought all the longer I could go at home was a couple, three minutes, when in reality I can go 2-3 times that long. And all of that at ISO 1600. Which gives me a lot more signal to work with.
BTW, I live a little away from the burbs (iirc in a yellow area) between 2 large cities.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Did anyone take a look at that stacks? What did you think?
--------------------
Arran
My main website
My blog
|
KevinUK
Vendor (DSLR-AstroMod)
Reged: 08/22/07
Posts: 720
Loc: N 51'53 W 00'25
|
|
Quote:
..Canon's Digital Photo Pro in the tools menu, there's a linear checkbox, S...
I cannot seem to locate this option?
I am using Canon Digital Photo Professional 3.3 but I think you mean DPP Diigital Photo Processing  right?, where is this utility located?
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
It's in Digital Photo Professional, was there in 3.3 and many previous versions. With a raw image file selected, pull down the View menu and select "Tool palette". The linear checkbox is on the tool palette below the histogram.
Changing the mode to linear will only affect the way the image looks when you use "Convert and save" to save the image in another format or when you use "Transfer to Photoshop" on the Tools menu, which saves the file as a TIF before sending it to Photoshop. The raw file will be unchanged, though it will look different in Digital Photo Professional (much darker) unless you open the tool palette again and turn of the linear checkbox.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
nickatnight
Carpal Tunnel
Reged: 04/14/08
Posts: 1751
Loc: Santa Clarita, CA (LA Suburbs)
|
|
Nice images and a super thread you got going here Arran. Very informative.
--------------------
nickatnight
Celestron CPC 9.25" GPS
Lumicon 2" LumiBrite Diagonal
Meade s5000 18mm UWA, Meade 6.7mm UWA,
Parks G.S-5 10mm, Celestrom 40mm Plossl
Parks 2x Barlow, Meade Tele-Extender
Celestron f6.3 Focal Reducer
Lumicon UHC Filter and Deep Sky Filter
Canon Rebel XTi (prime)
Canon Powershot G9 (afocal)
Canon EOS T-Ring and Parks T-Adapter
Lumicon Univ DigiCam Adapter
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Hi Arran,
Here's my attempt at the two 7x10 minute stacks. That's the dark sky image on the left. The LP image on the right has a pretty strong color cast which I didn't try to fix, due to the CLS filter, and most visible in the stars. I will take a shot at the 13 hour stack a little later.
As you can see the amount of detail in the two images is pretty similar, which is to be expected with the same exposure times, though I think you could do a little better at the LP site without the filter. The amount of noise in the LP shot is definitely higher. At 100% it's considerably more noticeable than in these reduced crops.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
Edited by DonR (04/27/09 07:33 PM)
|
bryansay
super member
Reged: 02/25/09
Posts: 111
Loc: Georgetown, TX
|
|
Don/others, can you recommend a specific book or two on Astro image Processing? I would prefer to buy on recommendations rather than title.
Thanks,
--------------------
Bryan Sayler
Celestron C9.25; Orion ED80
HEQ-5; Orion ST-80 w/starshoot
Nikon D80; Canon 40D Hap Griffin Mod
Empty Wallet 
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Hi Don
Thats a good job you have done on those. Just goes to show how much can be processed away with the right skills!
The dark sky image does look a lot cleaner though, but I guess once you go past a few hours of subs, the noise at a light polluted site will be more under control.
Looking forward to see if you have a go at the big stack!
Cheers
--------------------
Arran
My main website
My blog
|
scopemankit
super member
Reged: 01/27/09
Posts: 142
Loc: south africa
|
|
I too, took the liberty of working with your second frame. I upped gamma, upped dark and did an unsharp mask.(PSP5)
A lot of detail was captured in your image, squeeze it out!
--------------------
Chris
Cape Town
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Quote:
Don/others, can you recommend a specific book or two on Astro image Processing? I would prefer to buy on recommendations rather than title.
Thanks,
Hi Bryan
I have recently purchased these online tutorials and they are helping a great deal:
http://ip4ap.com/
Another book that is highly rated (I have not seen it myself though):
Photoshop Astronomy by R-Scott-Ireland
--------------------
Arran
My main website
My blog
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Hi Chris
The frames I showed were not fully processed, there is a lot more in there. They were just to show the difference in sky quality by doing the exact same basic stretch on each.
Download the 13 hour stack and see what you can make of that...my own attempt is half way through the thread.
Cheers
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Hi Arran,
I worked on your 13 hour stack, and it was a joy to work with. Very nice work capturing it. Here's what I came up with:
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Hi Don
Thats a great job! I definitely need to keep practising my processing skills. Are you able to post a full size uncompressed jpeg by any chance? So I can see it fully
Cheers
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Yep, here it is - I would have put it on my photo site, but people might think it was mine - nice for me but unfair to you!
BTW, I only use curves in Photoshop plus a couple of special manipulations using native Photoshop tools. The green channel, which has the best overall contrast, was copied to a separate layer and blended in Luminosity mode. The red channel, which has the cleanest background (due to the CLS filter I guess) was copied to a separate layer and blended in Overlay mode.
An appropriately mild noise reduction with something like NeatImage would clean up the little remaining noise in the faint outer reaches of the galaxy and around the brightest stars.
http://dandjreed.homedns.org/arran/13_hours_a_fullsize.jpg
Cheers,
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Thanks Don! Very nice process you have done there.
I have never tried this blending ideas, but I shall have a look at trying them out.
I recently checked out the www.ip4ap.com tutorials. There are some great things there for helping out the colors.
Also been attempting increasing the star colors recently, and more uses of various masks for sharpening etc.
So much to learn though!
Thanks again
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
You're welcome, Arran.
BTW, I still think you might have a better image without the CLS filter for this object. The total illumination for the galaxy would increase, and the shot noise would decrease accordingly. Without the CLS filter the sky background would probably be the noisiest area of the stacked image, but cleaning up the shadows would be easier than cleaning up the faint midtones.
You probably don't want to try another 13 hour project, but if you do I would suggest keeping the sub lengths the same without the CLS. The sky glow will be considerably brighter of course but there would still be plenty of dynamic range left.
By all means go to your dark sky site when you can, but when you can't, learn to make the most of it. For me the nearest dark site (blue zone) is 100 miles away and privately owned. The local astronomy club has access to that site, but their meetings are 35 miles away. There are some public areas that aren't quite so dark but substantially darker than my home site and as close as 40 or 50 miles. I will get around to visiting some of those, or better yet getting active in the astronomy club.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Yeah perhaps I should try imaging without the CLS. So far I have used it with every shot I have taken!
When I ever I go for a target, I generally get as many hours as I can. My last image was M81 and M82, and I think I got 13 hours on that one too.
At home my view is blocked in quite a few areas, so I dont have much choice in targets, which is why I tend to spend so long on the ones I like.
As you have shown, all is not lost with imaging in light polluted skies. I will still make the trip to dark skies whenever I get chance though. Looking up with the naked eye is far more entertaining there
--------------------
Arran
My main website
My blog
|
bryansay
super member
Reged: 02/25/09
Posts: 111
Loc: Georgetown, TX
|
|
Thanks Aza, sure appreciate this thread. The exchange between knowledgeable users has been very thought provoking. I too am in a LP area but plan to use filtering to maximize my available imaging time. I am encouraged by what I read here about processing LP out.
Clear dark skies,
--------------------
Bryan Sayler
Celestron C9.25; Orion ED80
HEQ-5; Orion ST-80 w/starshoot
Nikon D80; Canon 40D Hap Griffin Mod
Empty Wallet
|
Samir Kharusi
scholastic sledgehammer
Reged: 06/14/05
Posts: 988
Loc: Oman
|
|
I think there is much more to be gained from this exercise than has been done so far. Let's first set out "accepted" theoretical principles and check them out one by one:
1. Subexposure lengths have to be such that all the subs are in skyfog-statistics-limted regime. Since these were 10min each we just have to verify that the subs were long enough for the dark site. 10 minutes is long enough for a focal ratio of f9 or faster for the darkest sites on Earth, explained here. I am assuming that the focal ratio was faster than f9 and consequently all the subs were "long enough".
2. Theory subsequently states that the SNR (Signal to Noise Ratio) from the two stacks will be equivalent if the INTEGRATION times at each site are of the same ratio as the ratio of the skyfog at those sites. Aza estimates that the skyfog ratio was Stellar Mag 1, roughly. I.e. all one would need is a 2.5 hour integration time from the light polluted site to to deliver an "EQUIVALENT" SNR as at the light polluted site. Aza shot 12 hours. In theory the image from the light polluted site should have lower SNR! I do have a reservation here as regards the difference in skyfog. However, it is fairly easy to extract it from the subs themselves, and much more reliable than visual estimates.
In brief, PROVIDED the subs are longer than the minimal required, shooting longer subs should have little impact on the final SNR (in theory anyway, but I am open to a theoretical derivation that demonstrates otherwise).
The way forward to extract the max out of this beautiful set of captures:
Make an estimate of the actual skyfog at each location from typical subs so that we know what ratio of integration times we should compare, and then use a suitable stack to compare with the one-hour stack at the dark site. To measure the skyfog at each location, can you please convert one typical sub from each location as if it were a daytime photo (NOT using Linear!) and take a pixel reading from the background sky (or positions of the skyfog mountains on the histograms). That should give a ratio of the integration times required to match the SNRs from the two stacks. For 12:1 ratios in integration times, theory states that the skyfog can differ by 2.5 stellar Mags! and you still can end up with the same SNR.
--------------------
Bored? Peruse my website:
http://www.samirkharusi.net/
|
Samir Kharusi
scholastic sledgehammer
Reged: 06/14/05
Posts: 988
Loc: Oman
|
|
The use of a non-Linear conversion makes it possible to measure the absolute skyfog at the dark site if the camera were unmodded. But the ratio of the 2 skyfogs can also be extracted by using Linear conversions of the 2 subs, without any post-processing. Just pixel read-outs of the sky backgrounds should be sufficient. The use of the CLS filter would be measuring the effective skyfog at the polluted site, good enough for this "research".
--------------------
Bored? Peruse my website:
http://www.samirkharusi.net/
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Hi Samir
This stuff is all a little over my head
I can post a sample sub from each location, but not sure what you mean here:
"convert one typical sub from each location as if it were a daytime photo (NOT using Linear!)"
I would be interested to see what the optimum exposure times for each site are etc...
I have also been told before that 6-10 minute subs are the maximum you should go for with a DSLR, as anything over that length is not benficial as you start getting more noise compared to signal.
I used to shoot 20 minute subs, but after hearing this I dropped down to 10.
--------------------
Arran
My main website
My blog
|
gatsbyiv
member
Reged: 03/29/09
Posts: 67
Loc: Whitehouse Station, NJ
|
|
I agree with Bryan: this is a tremendously educational thread for those of us wrestling with LP, processing, and determining the proper sub times.
Don and Samir, please continue the discussion!
--------------------
-----------------------
Astro Telescopes 102ED
Astro Telescopes 80ED
Meade LXD75 Mount
Canon Rebel XSi
Just enough knowledge to be dangerous
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
I have also been told before that 6-10 minute subs are the maximum you should go for with a DSLR, as anything over that length is not benficial as you start getting more noise compared to signal.
I used to shoot 20 minute subs, but after hearing this I dropped down to 10.
Hi Arran,
It is not true that subs longer than 6-10 minutes with a DSLR cause more noise to creep in. In fact, the total SNR continues to increase with exposure time until your sub is completely white with an infinitely high SNR.
What does happen with longer subs, of course, is the sky background gets brighter. But the subject illumination always stays ahead of the sky glow with a digital camera, so the point at which exposure length becomes a problem depends entirely on your site, your equipment and your settings, e.g., light pollution, focal ratio, camera's dynamic range and ISO. If your sky and equipment allow it, longer is always better until the subject details you want to keep begin to saturate the detector.
If you consider only the signal to noise ratio, it would be best to shoot at the lowest possible ISO with the longest possible exposure times, until the detector won't hold your data any more. For practical reasons, though, you may need to limit the exposure times and/or raise the ISO in order to capture enough subs in the time you have available. That consideration, plus your mount's ability to track combined with your ability to guide, should be the basis for your decision.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Thanks Don.
Its looking good for me to get out and image this evening.
Based on the subs of mine you have seen, what settings do you suggest I try?
I will be imaging from my light polluted home, and tracking is not an issue, previously I was going for 20 min subs with no problems.
I will be imaging M51, and from your suggestions I wont be using the CLS filter. So what ISO and exposure length do you think I should try out? It will be with my F5 8" Newt.
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
I don't have a CLS filter Arran so I don't know just how much it reduces the signal. But from what I saw in your subs, you still have lots of room for data. I would start by trying at least the same length you use with the CLS filter, and evaluate the first sub using the Canon Digital Photo Professional software. Look at the position of the "hump" in the histogram - it represents the sky glow. Expose until it is close to the center of the histogram. You can push it past the center (to the right half) if you need to but I think it's better not to go that far if you don't have to, and I don't think you will have to.
I don't recall seeing the ISO you used for the subs you provided earlier. Keep in mind that while lower ISO/longer exposures are good, acquiring many subs is good too, so there's always a trade-off unless your time and patience are unlimited. If you have plenty of time (split over multiple nights if necessary and if the weather allows) and don't mind putting off the final results, going long and slow will pay off in the end.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Mike Clemens
Post Laureate
Reged: 11/26/05
Posts: 4278
Loc: Wasilla, Alaska 61N
|
|
(CCD theory)
(more discussion of theory)
--------------------
historius apochromaticus
|
Samir Kharusi
scholastic sledgehammer
Reged: 06/14/05
Posts: 988
Loc: Oman
|
|
I suggest that anyone who wishes to grasp the theoretical side of all this discussion please read and digest this article by Steve Cannistra , not by me, so I have no axe to grind or a point of view to push. It really discusses all the salient issues and concerns on a robust, theoretical basis. Many of the discussions on this board and other Groups tend to use and extend half-truths or half-comprehensions. Steve Cannistra is a well respected CCD imager and his paper has been "vetted" by the top amateur imagers on the SBIG board and CCD imaging technology boards. So it has a lot of street cred. I tried to distil its conclusions skirting the maths, to which most people seem allergic, and wrote this discussion paper. Before publishing it, I asked Steve Cannistra to review it, and he gave his OK. So I think it is reasonably robust. My conclusions and understanding of the mathematical interactions have enabled me to switch from autoguiding to completely unguided imaging.
Issues that still need addressing by field tests, and the two stacks available on this thread may indeed aid in resolving:
1. How much benefit is available for grasp by increasing "Minimal" length subs (1 minute at f2.8) to say, 2 or 3 minutes? We wish to get on Steve Cannistra's 90+% asymptote, not really interested with going from 90 to 95% if we have to extend the subs a factor 20x in order to do so. But if a factor 2x gets us from 70% to 90% then it is indeed of interest.
2. Once the subs are sufficiently long to put us at the 90+% stacking efficiency level, as they seem to be in both the stacks under discussion, then is it possible to field-verify that the SNR in the integration-time-equivalent stacks will really be equivalent, as theory predicts? Or are there other sources of noise that currently-accepted theories do not include?
I believe that the above stacks could answer 2, if Aza cooperates.
I have never come across ANY theoretical paper that advocates well-fill as a criterion for stacking efficiency. So trying to get the skyfog hump as far to the right, without clipping, does not seem to have theoretical backing. I do not say that it is not the "truth", just that current theories do not lead to such a conclusion. Longer and ever longer subs have little to do with the final result SNR, PROVIDED(!!!) each sub is longer than a certain minimal length, one that gets you into the Skyfog-Statistics-Limited-Regime. From Cannistra's theoretical derivations this is attained when the skyfog is 30+ times the Read Noise of your camera. The Read Noise of Canon DSLRs is often much lower than many astro CCDs or other DSLRs. It took me a lot of searching to finally understand why those of us using Canon DSLRs seem able to get away with subs often much shorter than even people using astroCCDs.
Aza: please convert one typical sub from each stack (using either Linear or non-Linear to a 16-bit tif) in Canon DDP and give me a pixel reading of the skyfog. If you do not follow what I mean, just crop a small patch of your converted tiffs at the same region within the frames and email the two to me. We can continue from there.
--------------------
Bored? Peruse my website:
http://www.samirkharusi.net/
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Hi Samir
I have just emailed you
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Hi Samir,
I think you misunderstood the reason for my recommendation to watch the position of the sky glow hump on the histogram and the saturation level of the sensor.
There is a very common misconception among amateur DSLR astrophotographers that the presence of substantial light pollution prohibits photographing faint objects, and the observable fact that the sky glow hump rapidly progresses well past separation from the left end of the histogram is used to "substantiate" this misconception. Part of the reason for this is older texts that refer to the "sky fog limit", a film related phenomenom which doesn't exist in digital astrophotography. The fact is with digital sensors, as opposed to film, the signal from the subject always remains above the sky glow level. The effect of the sky glow is to push the baseline higher, which reduces the dynamic range left for the subject and makes post-processing more difficult. It is still true, however, that longer exposures increase both the SNR and the signal, and it is the latter which some theories seem to ignore.
Forgetting about noise for a moment, there is a minimum signal level required in order to produce a reasonable image of a DSO, and that signal level is largely independent of sky glow level. In addition, the only upper limit to the signal level required for an optimal image is the full well capacity of the sensor. There is indeed an upper limit to exposure length at which the SNR of the sky glow begins to level out, and longer exposures produce dimishing returns if only the sky glow's SNR is considered, but you must also consider the SNR of the subject! It continues to increase with exposure length until the sensor can no longer hold the data.
So IMHO any theory that only considers the SNR of the sky glow falls short when it comes to predicting the optimal exposure time. The reality is considerably more complex, especially when photographing from areas with substantial sky glow. You must consider how much signal from the subject is required to produce the image you desire, and you must consider how much dynamic range you need left, after clipping out the sky glow, to stretch the remaining signal. These considerations cannot be considered without taking into account the nature of the subject.
I have adopted an approach based on my experience at my site, which is fairly heavily light polluted. With the faintest and most diffuse objects I try to photograph, I have found that pushing the exposure until the sky glow hump is about 2/3 of the way to the right produces optimal results. To many people this sounds outrageous, but the fact is I am only using at the most about 20% of the sensor's range for the sky glow in that situation. The reason that the hump in the histogram is so far to the right is that the histogram in the software I use (Canon DPP, Photoshop) is displayed using a logarithmic curve. With Canon DPP, you can switch to a linear histogram, and doing so gives you a better idea of how far to the right the hump really is. And using Iris software you can get an accurate measurement.
I downloaded and processed Arran's data and my observation was that his light pollution level is considerably lower than mine, though in fact I don't have a CLS filter so I can't quantify it precisely. That was the basis for my recommendations to him regarding monitoring the sky glow hump and pushing it near the middle of the logarithmic range but not past.
I am still interested in Steve Cannistra's theory and your elaborations on it. I downloaded his spreadsheet, but have not been able to use it yet. The issue is that Steve's spreadsheet only contains gain and noise data for some of the SBIG CCD's, and I use a Canon DSLR. I evaluated the sky ADU from some of my images and plugged the numbers in, selecting various SBIG CCD's as the "Camera Used", and in all cases the spreadsheet calculated my subexposure needed for 90% F as negative numbers. Can you (or anyone reading this) provide appropriate values for gain and readout noise to use with a Canon 350D?
Cheers,
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
groz
Pooh-Bah
Reged: 03/14/07
Posts: 1078
Loc: Duncan, BC
|
|
Quote:
1. Subexposure lengths have to be such that all the subs are in skyfog-statistics-limted regime.
This is the part I have a problem grasping, so bear with me samir, while I ask a long question on this particular part.
I'll use an example of a shot I recently did, M51, from our back deck location. Part of my confusion may indeed come from the fact our location seems to be a lot better than locations you folks are always referring to. On a clear moonless night, when seeing is decent, mag 5 stars are an 'easy' catch naked eye, and lots of visible milky way is very normal here when there isn't a moon in the sky. Sometimes I wonder if maybe the issue is, I dont have enough sky glow for the 'sky glow limits' to come into play before saturation starts happening on bright parts of the target.
When I shot M51 a couple weeks ago, using the C-8 with the 0.63 reducer and the canon 350D at prime, I started by taking a few shots and looking at histograms. 3 minutes gave me a histogram with the 'sky fog' bump at about 25% from the left side, nicely displaced from the origin (about 10%). 5 minute exposure got the sky fog bump about 40% over, and 6 minute exposure got it about 50%. Whenever I read your stuff regarding 'sky fog limits', it suggests that for this location, the 6 minute sub would be appropriate.
But, the problem I then run into with the 6 minute shot is within the galaxy cores, they are completely overexposed, and it's not possible to pull any gradients out of the resulting stack. Essentially, the galaxy core has come close enough to saturating the sensor, there is no real gradient in this area anymore. This is not the case with the 3 minute sub, there are good gradients in the bright galaxy cores, and after I stack enough of them (60+) the background noise is reduced enough I can pull good gradients out of the dim galaxy arms.
Looking in more detail at what's going on with the histogram, the sky fog hump is more or less a normal distribution around a point, which makes sense, one would expect the sky fog to have a more or less normal variance. That big bump in the histogram represents all of the background, but, the pixels of interest (ie the target) are actually those in the histogram to the right of the bump.
so the crux of my question. Why is there an incessant desire to focus on the exposure times for the background of the image, without taking into account the the actual target (foreground) ? In the end, it is the foreground we want in the image. If I do exposures here based on the position of the sky fog hump, my actual target ends up overexposed with my camera / telescope combination.
I've read a lot on your website, and the math all does indeed make sense to me, but, it also seems to overlook one detail. If we base the exposure length entirely on the background glow levels, and ignore the issue of bright areas within the target, we do end up with a perfectly exposed background, and an overexposed target.
So, after figuring out all the limits based on the 'sky fog limits', doesn't it make sense to revisit the actual target, and confirm that the target itself is not overexposed, and reduce exposure time down from there, to keep the target data within the range of the sensor ?
What am I missing here ??
|
Samir Kharusi
scholastic sledgehammer
Reged: 06/14/05
Posts: 988
Loc: Oman
|
|
Wow! Each of the questions posed above requires a lengthy explanation. OK, I will take the simplest one first, galaxy-core over-exposures. Most imagers consider this as a non issue, as anyone who has shot M42 (the Orion Nebula) will attest. Just use short exposures for the core and blend them in later. The problem is ALWAYS those faint tendrils, so all the discussion tends to stay with them, and the way to get good SNR there is to pay attention to skyfog. With DSLRs we can always use a lower ISO to extend the dynamic range for longer exposures, Eg at my polluted home base I normally use ISO 200 to 400. At a dark site I normally use ISO 1600.
--------------------
Bored? Peruse my website:
http://www.samirkharusi.net/
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Here is a single 10 minute sub from last night, without the CLS filter.
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Arran, your 10 minute sub is well within the maximum skyglow limits I would use to evaluate a single sub, based on experience rather than mathematics. What was the ISO? If more than 200, then your sky glow is significantly less than mine, as I suspected.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Hi Don
This was ISO800. So with these conditions, you would shoot 10 minute subs?!?
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Hi Don
This was ISO800. So with these conditions, you would shoot 10 minute subs?!?
Yes, Arran, I would start with 10 minute subs, and based on the results even consider extending that. You could also consider ISO1600, but with my 350D at least that generally isn't productive.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Hi groz,
Regarding the burned out core, Samir's advice to shoot another set of shorter subs for the core exposure may be the only way to solve the problem, but I wouldn't jump to that conclusion yet.
One of the reasons that I prefer Iris software over DSS is that Iris has a few very useful post-processing features that are missing or poorly implemented in DSS. I don't know what software you use for post-processing, but my experience with DSS is that you should forego all post-processing until you export the stacked results into Photoshop.
And Photoshop even falls short of Iris in a couple of respects relevant to DSO photography. Those are the sky gradient removal and the non-linear stretching capabilities.
While Photoshop has both capabilites they are much easier to implement in Iris IMHO. The SUBSKY command in Iris takes a little tweaking sometimes but generally produces very good results. Even more important though is the non-linear hyperbolic arc-sin stretching that you can do in Iris. In my experience, with a little tweaking, you can stretch the contrast with Iris' ASINH command to bring out the faintest details you have recorded without blowing away anything but the totally saturated highlights. I am routinely surprised when looking at a stacked result before and after applying the ASINH command by how much of the highlights, which appear to be totally saturated, can actually be preserved. The same results could theoretically be achieved with Curves in Photoshop, but in practice doing so is nearly impossible.
Perhaps other astro-specific applications than Iris have similar capablilities but I haven't found one yet.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Samir Kharusi
scholastic sledgehammer
Reged: 06/14/05
Posts: 988
Loc: Oman
|
|
Quote:
Hi Samir,
So IMHO any theory that only considers the SNR of the sky glow falls short when it comes to predicting the optimal exposure time. The reality is considerably more complex, especially when photographing from areas with substantial sky glow.
Cheers,
Actually, the theories are chasing the SNR of the target object (after the skyfog has been subtracted out) and nobody gives a hoot what is the SNR of the skyfog. That's just enroute to the promised land. The concepts are more difficult to express in words than in maths and Cannistra's paper is the most eloquent I have come across so far. OK, you have skyfog signal and skyfog noise, generally both much larger than target-object signal and target-object noise. Simply by increasing Integration time we can eventually narrow the statistics enough to subtract (or disentangle) the object signal+noise from the skyfog signal+noise. The battle is to use sufficient Integration time. If one uses a long-enough single exposure, there is no further quibble. But if we wish to use shorter sub-exposures and then stack them we have to make sure that each sub is not messed up by a noise source that is so random and unpredictable that each sub simply adds up the noise, rather than narrowing things down with the sqrt rule. Camera Read Noise is one such noise source. Hence we want to operate as much as possible, away from Read Noise. From Cannistra's derivations one would get 90+% stacking efficiency if each sub puts the (larger) skyfog+noise at 30x the Read Noise. This is what is meant by the Skyfog-Statistics-Limited Regime. His maths do NOT impose a minimum length of sub per se. No matter how short the subs are, they will always stack to give an image, but the stacking efficiency (compared to a single long exposure)drops dramatically below 90% very quickly if we allow the Read Noise to interfere with the skyfog signal+noise.
Your experience and gut feel says that to attain that 90+% stacking efficiency when at a site with high skyfog you need to shoot subs much longer than the minimal 30x Read Noise. I cannot say that your conclusion is invalid, just that the maths does not seem to indicate that. Perhaps you can dig into Cannistra's maths and point out either where this effect is seen in the maths, or where his basic assumptions have been negligent. I am unable to do so.
Read Noise in Canon DSLRs seems to be about half a percent to a third of a percent of the Histogram X-axis. You can just shoot a dark frame at 1/8000sec at ISO 1600 and check it out for your own camera, or do it more scientifically by downloading the latest beta version of ImagesPlus 3.80, or use Iris. Hence the usual adage of putting the skyfog at around 25% should be >30x Read Noise, i.e. put us into the 90+% stacking efficiency regime. But your experience basically says, "No, the skyfog has to be much more than 25% of the histogram in each sub, in fact around 75+% to achieve Cannistra's magical 90+% stacking efficiency". This statement could be valid if the Read Noise of the camera is indeed around 2 to 3% of the histogram, rather than <<1%. Perhaps Canon are doing some subtraction inside the camera? Possibly and it should be possible to verify that using Iris or ImagesPlus 3.80. Seems that Nikon or Pentax or somebody does the subtracting in-camera. I do not now recall which one. ImagesPlus 3.80 specifically tries to get the really, really Raw files in an attempt to get rid of noise bands.
James Clark at Clark Vision has published a lot of basic data on DSLRs. That could help in identifying what is going on.
Skyfog being "high" is relative, of course. The skyfog through an f2 OTA is always very, very high compared to the skyfog through an f10 OTA at the same site. So if your conclusions are valid as regards exposing to the right, then the conclusions should be valid everywhere. Methinks.
The only way I can think of to confirm or refute your experience-based conclusion/deduction would be to shoot at the same location and sky conditions, subs that put the skyfog at, say, 20%, 40% and 80% of the histogram X-axis, adding to identical integration times and compare the results, with each set properly calibrated with bias, darks (and flats?). If the 80% obviously wins, then we have to come up with a theoretical basis that makes sense (eg a Read Noise as high as 3%) or some as yet not thought-through cause.
The concept of putting the skyfog well above the Read Noise so as to establish that each sub is in the SSL regime is well established among astroCCD imagers. What is not established definitively is the value of Read Noise in any given DSLR, what secret in-camera subtractions are going on, etc. IMHO it would be worthwhile to dig deeper into these technicalities before inventing new theories.
--------------------
Bored? Peruse my website:
http://www.samirkharusi.net/
|
Samir Kharusi
scholastic sledgehammer
Reged: 06/14/05
Posts: 988
Loc: Oman
|
|
Quote:
Hi groz,
Regarding the burned out core, Samir's advice to shoot another set of shorter subs for the core exposure may be the only way to solve the problem, but I wouldn't jump to that conclusion yet.
Perhaps other astro-specific applications than Iris have similar capablilities but I haven't found one yet.
I was assuming that he knows how to use "Curves", but perhaps I was assuming too much. In my experience galaxy cores (as opposed to M42) ever causes a major issue. ImagesPlus does provide these more esoteric curve functions, eg by using DDP. If one is trying to use a daytime image editor, eg Photoshop, the trick is to do the stretching in multiple stages, rather than in one go. Each stage retains the top end of your initial histogram so that no star or galaxy core reaches saturation compared to the intial stack, and you stretch only the lower values. By using 4 or 5 stages one can approach an arcsin stretch (or any other stretch) quite well.
--------------------
Bored? Peruse my website:
http://www.samirkharusi.net/
|
Tonk
Post Laureate
Reged: 08/19/04
Posts: 4358
Loc: Leeds, UK, 54N
|
|
Quote:
Perhaps other astro-specific applications than Iris have similar capablilities but I haven't found one yet.
How about DDP in Images Plus and MaxImDL. I use the IP version of DDP with the sharpening filter turned off
--------------------
Televue 85, GM-8/Gemini, Canon 40D (unmodded), Canon 450D (modded w/Astronomiks clip-ins - UV/IR, OWB)
Coronado SM60/BF10, Baader Herschel Wedge
Leeds Sky Clock Ripon Sky Clock
|
Tonk
Post Laureate
Reged: 08/19/04
Posts: 4358
Loc: Leeds, UK, 54N
|
|
Quote:
If one is trying to use a daytime image editor, eg Photoshop, the trick is to do the stretching in multiple stages, rather than in one go. Each stage retains the top end of your initial histogram so that no star or galaxy core reaches saturation compared to the intial stack, and you stretch only the lower values. By using 4 or 5 stages one can approach an arcsin stretch (or any other stretch) quite well.
Exactly so - I use a gentle curves/levels iterative pair after doing IP DDP and discussed this on anotther thread yesterday when discouraging folk from processing astro images with brightness and contrast operations
--------------------
Televue 85, GM-8/Gemini, Canon 40D (unmodded), Canon 450D (modded w/Astronomiks clip-ins - UV/IR, OWB)
Coronado SM60/BF10, Baader Herschel Wedge
Leeds Sky Clock Ripon Sky Clock
|
Tonk
Post Laureate
Reged: 08/19/04
Posts: 4358
Loc: Leeds, UK, 54N
|
|
I've not seen the Canistra paper before, but I have been using the John Smith paper as a guide sinse the paper first appeared
http://www.hiddenloft.com/notes/SubExposures.pdf
I too have used this to work unguided!
--------------------
Televue 85, GM-8/Gemini, Canon 40D (unmodded), Canon 450D (modded w/Astronomiks clip-ins - UV/IR, OWB)
Coronado SM60/BF10, Baader Herschel Wedge
Leeds Sky Clock Ripon Sky Clock
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Hi Samir,
Quote:
Actually, the theories are chasing the SNR of the target object (after the skyfog has been subtracted out) and nobody gives a hoot what is the SNR of the skyfog. That's just enroute to the promised land.
I don't see how you can possibly consider the SNR of the target object without actually measuring it. The only measurements I see in Cannistra's paper are for the sky glow.
Let's consider two extremes. The moon, while not a DSO, is still an extended object with a wide dynamic range. The calculation of optimal exposure time based on SNR of the sky glow will completely fail to predict a valid exposure time for photographing the moon. In an earlier response you cited overexposed galaxy cores and the Trapezium as examples where it is necessary to take a separate set of subs in order to properly expose those areas of your subject. But suppose that the Trapezium IS your subject, and you care nothing about the rest of M42. Clearly if you don't consider the illumination of the subject but only the sky glow, you will totally fail.
On the other extreme, consider a very distant and faint galaxy near the limit of detection with your equipment and location. If you expose based on the sky glow, and only capture a few photons in some of the subs, that signal will be treated as noise in the stacking process and smoothed out, erasing your subject. You must capture enough photons in each sub, or at least in the majority of subs, to define the subject. And if that faint galaxy has some texture that you hope to capture, you must capture even more signal in each sub to prevent the stacking process from wiping out the subject details. In this case again, exposing to "optimize" the sky glow SNR will fail to produce the desired results.
Quote:
His maths do NOT impose a minimum length of sub per se.
Then perhaps the major objection I have is the implication that the calculated subexposure lengths are "optimal". In fact they might be considered "maximal" in some cases and "minimal" in others, depending on the subject. This makes them, IMHO, interesting at best and misleading at worst.
Quote:
Read Noise in Canon DSLRs seems to be about half a percent to a third of a percent of the Histogram X-axis. You can just shoot a dark frame at 1/8000sec at ISO 1600 and check it out for your own camera, or do it more scientifically by downloading the latest beta version of ImagesPlus 3.80, or use Iris.
Yes, I can calculate the read noise of my DSLR in Iris and have done that. But to use Cannistra's spreadsheet I need to enter the camera gain in charge per ADU, and I don't know how to calculate that.
Quote:
The only way I can think of to confirm or refute your experience-based conclusion/deduction would be to shoot at the same location and sky conditions, subs that put the skyfog at, say, 20%, 40% and 80% of the histogram X-axis, adding to identical integration times and compare the results, with each set properly calibrated with bias, darks (and flats?). If the 80% obviously wins, then we have to come up with a theoretical basis that makes sense (eg a Read Noise as high as 3%) or some as yet not thought-through cause.
In my experience about 60% to 70% wins for faint DSO's, and well below 50% for brighter DSO's with less contrast. But these figures are based on the log scale histogram, and translate to perhaps 20% for faint DSO's and less than 10% for brighter DSO's, using the linear scale. Once again, you can't come up with a figure that works, IMHO, without considering the subject.
However I think our primary disagreement is related to your last sentence from the paragraph above. I do not feel any compulsion to come up with a theory that makes sense, as long as I know what is needed in my subs.
Consider professional photographers, whether engaged in astrophotography or conventional daylight photography. They know from experience how to shoot a scene, and how to modify their exposures based on the results they are seeking. I've yet to see one pull out a calculator or resort to a spreadsheet before setting the camera, but I have seen them try different exposure values and evaluate the results on the fly. In the old film days, it was sometimes necessary to shoot many exposures using different camera settings since it wasn't possible to review the results instantly and adjust accordingly.
With beginning astrophotographers I recommend looking at what others with more experience do, and experimenting, since equipment and imaging sites vary. I recommend strongly against blindly following theoretical calculations. However I do encourage you and others to continue pursuing the theoretical aspects of exposure because for some (many?) of us understanding the technical details is very helpful and interesting.
Regards,
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Quote:
Hi groz,
Regarding the burned out core, Samir's advice to shoot another set of shorter subs for the core exposure may be the only way to solve the problem, but I wouldn't jump to that conclusion yet.
Perhaps other astro-specific applications than Iris have similar capablilities but I haven't found one yet.
I was assuming that he knows how to use "Curves", but perhaps I was assuming too much. In my experience galaxy cores (as opposed to M42) ever causes a major issue. ImagesPlus does provide these more esoteric curve functions, eg by using DDP. If one is trying to use a daytime image editor, eg Photoshop, the trick is to do the stretching in multiple stages, rather than in one go. Each stage retains the top end of your initial histogram so that no star or galaxy core reaches saturation compared to the intial stack, and you stretch only the lower values. By using 4 or 5 stages one can approach an arcsin stretch (or any other stretch) quite well.
I'm quite sure groz knows how to use Curves in Photoshop and didn't mean to imply that he doesn't. When I use Curves in Photoshop I do exactly as you suggest, using them in small increments, frequently taking 20 or more steps to implement a relatively small adjustment. However I still have never been able to achieve anything close to what I can achieve with the ASINH command in Iris with just one pass. I could try harder I guess, but why should I when Iris is free and extremely effective?
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
Quote:
Hi Samir,
Quote:
Read Noise in Canon DSLRs seems to be about half a percent to a third of a percent of the Histogram X-axis. You can just shoot a dark frame at 1/8000sec at ISO 1600 and check it out for your own camera, or do it more scientifically by downloading the latest beta version of ImagesPlus 3.80, or use Iris.
Yes, I can calculate the read noise of my DSLR in Iris and have done that. But to use Cannistra's spreadsheet I need to enter the camera gain in charge per ADU, and I don't know how to calculate that.
Don, There's some great info here and here including ADU and Unity gain for canon dslr's.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Quote:
Perhaps other astro-specific applications than Iris have similar capablilities but I haven't found one yet.
How about DDP in Images Plus and MaxImDL. I use the IP version of DDP with the sharpening filter turned off
I haven't used Images Plus or MaximDL except for brief trial periods, but it's good to know that there are effective alternatives to Iris, should I ever need to switch for any reason.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Quote:
Hi Samir,
Quote:
Read Noise in Canon DSLRs seems to be about half a percent to a third of a percent of the Histogram X-axis. You can just shoot a dark frame at 1/8000sec at ISO 1600 and check it out for your own camera, or do it more scientifically by downloading the latest beta version of ImagesPlus 3.80, or use Iris.
Yes, I can calculate the read noise of my DSLR in Iris and have done that. But to use Cannistra's spreadsheet I need to enter the camera gain in charge per ADU, and I don't know how to calculate that.
Don, There's some great info here and here including ADU and Unity gain for canon dslr's.
Thanks, Mike, that's what I was looking for!
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
justabob
Carpal Tunnel
Reged: 05/05/07
Posts: 1685
|
|
Quote:
You're welcome, Arran.
BTW, I still think you might have a better image without the CLS filter for this object.
Don, I have never used the cls filter but untill I started using the idas filter from my mag 4 home site I stopped imaging rgb altogether. I was able to double my sub exposure times while achieving the same histogram. I find it improves not only nebulas but broad band targets as well.
--------------------
http://www.pbase.com/rkn/astro&page=all
Vixen Sphinx SXW
Meade sn6
Canon EF 200mm f/2.8L II USM Lens
Hutech 1000d
Self modded 350d
ST8300c on order
DSI PRO II
Bob
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Quote:
You're welcome, Arran.
BTW, I still think you might have a better image without the CLS filter for this object.
Don, I have never used the cls filter but untill I started using the idas filter from my mag 4 home site I stopped imaging rgb altogether. I was able to double my sub exposure times while achieving the same histogram. I find it improves not only nebulas but broad band targets as well.
I don't doubt it, Bob. Everyone's situation is a little different, which is why I try to resist broad generalizations.
Looking at the sub Arran posted last, and comparing it to the stack he posted earlier with the same exposure time but using the CLS, it appears to me that he has much more useful data without the CLS, and plenty of dynamic range left for post processing after he stacks a set of those subs. It's obvious from his data that his LP is relatively mild.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Don, due to the weather I did not manage to get any more subs, so I cannot do any testing of stacked images.
But looking at my image, I wouldnt know how to start processing it! When I use the CLS it all seems a lot easier, and then at the dark site it seems a whole more easier still! How would I start with the processing of this image?
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Don, due to the weather I did not manage to get any more subs, so I cannot do any testing of stacked images.
But looking at my image, I wouldnt know how to start processing it! When I use the CLS it all seems a lot easier, and then at the dark site it seems a whole more easier still! How would I start with the processing of this image?
Hi Arran,
Obviously you will start with adding more subs, but you already know that!
After you calibrate (subtracting darks, dividing by flats), register (aligning the stars) and stack, you should end up with an image that looks a lot like the sub but is much less noisy at the pixel level.
My workflow then would be to remove any remaining sky gradient, then adjust the white balance, then stretch. That's way too general, though, so this evening I will do a quick pass on that sub and post the results along with the steps I used. The results will be noisy of course, and there will likely still be some vignetting, unless you can post a flat field with the same dimensions as that sub.
Cheers,
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Thanks Don.
Well as I only took one sub before clouds so I didnt bother with any flats. I have some from other nights, but I guess the focus is slighty off on those.
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
That's OK, I'll synthesize a flat. It's better not to use flats from a different night unless you know everything (focus, camera rotation, collimation) is identical.
Edited by DonR (04/30/09 04:30 PM)
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Here is the single 10 minute tif:
http://www.mediafire.com/?myzdymgnt5t
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Hi Arran,
How did you create the TIF? Did you by chance have a custom white balance setting in the camera based on the CLS filter? The TIF has a very odd color balance, not just in the sky background but in the galaxy and the stars too, that I can't get rid of. Can you post the RAW file?
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
Samir Kharusi
scholastic sledgehammer
Reged: 06/14/05
Posts: 988
Loc: Oman
|
|
Quote:
Hi Samir,
Once again, you can't come up with a figure that works, IMHO, without considering the subject.
However I think our primary disagreement is related to your last sentence from the paragraph above. I do not feel any compulsion to come up with a theory that makes sense, as long as I know what is needed in my subs.
Regards,
Pity, because the way human knowledge expands is by the tiny, incremental understandings proferred by millions of individuals. In our age, we "understand" something when we can write out an equation that specifies the behavior under generalised conditions. Later on, we find that that equation is invalid under certain circumstances (like this placement of the skyfog histogram under bright skyfog). The equation then has to be modified, etc. As for the subject of the picture, I think most of us would consider that the subject of interest is invariably the faintest tendril of any nebula or galaxy. If we can coax that out, then we know how to coax out the brighter parts. Pity that these discussions tend to get passionate. Is it because, like politics, economics, social sciences, we are unable to write down an equation that represents the true state of affairs and how all the parameters of interest interact? 
--------------------
Bored? Peruse my website:
http://www.samirkharusi.net/
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Hi Don
The camera white balance is set to daylight.
That file I posted is simply the captured fit converted to tif.
Captured and converted by Nebulosity.
If I use DSS to stack, I choose RGB background calibration to fix that.
Or if I use Neb to stack, I adjust color background (offset).
Later on I will upload a tif that has been balanced by Nebulosity.
--------------------
Arran
My main website
My blog
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Hmmm Nebulosity seems to be having trouble balancing the colors.
--------------------
Arran
My main website
My blog
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
Long thread... sheez
Based on the article I'm writing, it would seem that one shouldn't be concerned with how far up the histogram the skyglow gets. Instead, the max exposure time should be that which is just before saturation of the highlights you wish to preserve. A higher skyglow will cause this point to come earlier. You don't need to expose this long of course, and Steve Cannistra's article does a good job of examining what the "minimum" exposure time should be. The article I'm writing hopefully will make a valueable contribution here.
Another large downside to skyglow that I haven't seen explicitly mentioned in this thread (if it's here than I missed it) is that the noise it generates combines with the noise from your target, reducing the overall max SNR you can acheive in a single subexposure.
To me, the answer to the question "how long should I expose for" would generally be "as long as you can while keeping the target steady and without overexposing the highlights you want to keep". Of course, there may be some limit of practicality when dealing with a faint target in really dark skies, and there is a point of diminishing returns.
--------------------
-Arkalius
11" Celestron SCT on Orion Atlas EQ-G
Celestron 100ED Refractor
8" Zhumell Dobsonian Reflector
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
Quote:
Long thread... sheez
.....
Another large downside to skyglow that I haven't seen explicitly mentioned in this thread (if it's here than I missed it) is that the noise it generates combines with the noise from your target, reducing the overall max SNR you can acheive in a single subexposure.
In a single exposure yes, but as your sampling rate increases I'm not so sure.
I think maybe what we're doing is becoming more applied digital signal processing than photography, and not that that should scare anyone away, but we rarely talk about sampling rates, and duty cycle, and how do we detect a single photon than arrives once every 10 seconds, and how is that different than detecting a photon that arrives once every 10 minutes?
Quote:
To me, the answer to the question "how long should I expose for" would generally be "as long as you can while keeping the target steady and without overexposing the highlights you want to keep". Of course, there may be some limit of practicality when dealing with a faint target in really dark skies, and there is a point of diminishing returns.
Agreed!
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
I have reprocessed my different stacks of data so many times now, that I have come to a decision on my future imaging!
I can clearly see that the data from the dark site is CONSIDERABLY better than I get at home.
Just one hours of data is so much better than many hours I get at home, it is well worth travelling for it.
The differences being I can stretch to get the faint detail with out causing a ton of noise to be introduced. And also the colors are so much better.
And that is with only one hours of data VS many hours.
The real test will be when I can get 4+ hours at my dark site. At that point I will truly know what the differences in sky quality will yield.
From my home I can produce very nice images, but I guess I am a perfectionist, and I am never truly happy with them. Travelling will enable me to get the images that I have always wanted.
Time will tell though, I need more data to be totally sure about this!
--------------------
Arran
My main website
My blog
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Hi Samir,
Quote:
Pity, because the way human knowledge expands is by the tiny, incremental understandings proferred by millions of individuals.
I couldn't agree more, but observations and measurements are at least as important as theories in the process of expanding knowledge. Typically observations come first, then theories to explain the observations, then measurements to confirm the theories. If the theories don't conform to the facts they need to be revised or abandoned.
The work that you, Cannistra and others have done is important, but in some cases I think it has been mis-characterized. You can obtain a figure that represents the minimum exposure time needed to reach the point on the sky glow SNR curve where longer exposures produce little improvement in the sky glow SNR, and that's good. But that point does not necessarily represent the optimal exposure time, because it doesn't consider the subject's brightness.
The way I estimate the exposure time needed with my equipment and in my location for an object I haven't attempted to photograph previously is to make a guess, based on experience, and then make some trial exposures. I examine the trial exposures closely, with very strong stretching if necessary, to determine if I am capturing any signal from the faintest parts of the subject that I desire to capture, disregarding the noise, and using a linear gamma if it helps the visualization. If I can't see any detail at all I know I must expose longer. When I do begin to see the detail I'm after (again disregarding the noise), I examine the position of the sky glow hump on the histogram and make a judgement (based on my experience) about whether I will have enough range left above the sky glow to produce a decent image. Sometimes the answer is "No", so I don't waste time trying to photograph that object from my light polluted site. If the answer is "Yes", then I determine if there is a reduced ISO and correspondingly increased exposure time that will allow me to achieve a better compromise between total number of subs and exposure time per sub. This last step is important because even if I already know the optimal exposure time to minimize noise in the sky glow, the SNR of the subject is a different matter. Once you get past the read noise, shot noise becomes the major contributor to noise in the subject, and the SNR due to shot noise increases with illumination, which increases with exposure time - and the SNR of the subject relative to shot noise continues to increase with added exposure time, with no upper limit short of infinity.
If someone can come up with calculations to make that process less subjective and to eliminate the need for experience, it would be very beneficial. But there are so many objects with such a wide range of brightness and contrast, the task seems quite formidable. In the mean time, I'll be taking pictures whenever I can.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
groz
Pooh-Bah
Reged: 03/14/07
Posts: 1078
Loc: Duncan, BC
|
|
Quote:
As for the subject of the picture, I think most of us would consider that the subject of interest is invariably the faintest tendril of any nebula or galaxy.
hehe, BINGO. I think this just goes to prove, different opinions on various things come about from DIFFERENT OBJECTIVES. In your case, the objective is the very faint details hiding down near the 'sky fog'.
On my wall, I have a poster, it's a hubble shot of M51, came in a skynews mag a couple years ago. When I'm processing my M51 stack, I'll often look at that poster, and compare to what I see on my screen, keeping in mind that I have neither the resolution, nor the clarity (seeing plays a big part here) of the hubble shot. Bottom line, my C8 + canon 350 is NOT a million dollar sensor strapped into a billion dollar telescope, mounted in a trillion dollar location.
But, the thing that really sticks out to me, when looking at that poster, the amount of detail in the vicinity of the galaxy core, as well as the detail in the faint tendrils. So, in my attempt to produce that kind of stuff with my modest gear, I _know in advance_ there's limitations I have to deal with.
a) If I expose to long, the gradients in the core start to wash out.
b) If I expose to short, I dont take advantage of the full range of the sensor.
c) I have a camera with relatively high read noise compared to dedicated astro equipment, so I need to take lots of frames to deal with that statistically.
d) The 12 bit a/d in my camera will _never_ achieve the detail of equipment with higher bit counts in the a/d, but I can indeed fake it down the road by using hdr techniques.
So, everything becomes a compromise, and, I try to adjust the compromise to reach _my_ goals, which turn out to be different than your goals. You are after the maximum detail in the faint stuff, prepared to sacrifice the brighter cores. I'm after the maximum detail in the cores, and prepared to sacrifice a bit on the dim end. You have higher end camera equipment (and telescopes), so you can push envelopes out of reach with my equipment.
And this is likely why discussions become passionate, folks are _assuming_ that everybody has the same objectives. In the end, i think we are all applying the same theory in general, but, working toward different goals, so we end up treating the equipment limitations differently.
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
Quote:
c) I have a camera with relatively high read noise compared to dedicated astro equipment, so I need to take lots of frames to deal with that statistically.
Actually, based on the math, it is the length of the subexposure, not the number of them, that has the greatest effect on read noise contribution.
Quote:
d) The 12 bit a/d in my camera will _never_ achieve the detail of equipment with higher bit counts in the a/d, but I can indeed fake it down the road by using hdr techniques.
Actually... the lower bit depth essentially amounts to a greater level of quantization noise, and, so long as you work with a program that will give you 16 bits of depth, this noise can be combatted the same way other noise can, with more integration time. At least, that's how it seems to me. It's hard to prove it algebraicly because quantization noise is hard to quanitfy statistically.
--------------------
-Arkalius
11" Celestron SCT on Orion Atlas EQ-G
Celestron 100ED Refractor
8" Zhumell Dobsonian Reflector
|
denjo
professor emeritus
Reged: 11/25/05
Posts: 506
Loc: Creve Coeur, IL
|
|
Arran,
I took a shot at processing your 13hr one. I have never processed a photo with anywhere near that kind of time in it and all I can say is WOW!!!
Now I know why you guys put so much time in your shots. That thing has tones of information in it. More than I know how to bring out and it was a joy to work on, THANKS for posting it.
My version has more blue than most people like but I go mostly by how it looks to my eye than whether or not its evenly balanced color.
Thanks again, this was fun.
--------------------
Coopatory Observatory
Astro Tech 111 APO
Atlas Mount
W/O 80mm EDII F/6.8
Orion Starshooter Guider
Black Knight Binoviewer
Canon 20Da
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Thanks Denjo, you have done a great job on the image!
I love to see how each person interprets the data
--------------------
Arran
My main website
My blog
|
Bart Declercq
sage
Reged: 01/21/07
Posts: 200
Loc: Haaltert, Belgium
|
|
I've processed your 13 hour M101 tiff, using Photoshop CS2 and Carboni's tools, I include two versions, one of the full image, but also one slightly larger, as I feel your image deserves the larger image scale.
Mostly playing with levels, using Shadows/Highlights and finishing touches using Curves - I removed a slight color gradient with Carboni's tools, then finally some sharpening, denoising and a high-pass filter overlay.
|
Bart Declercq
sage
Reged: 01/21/07
Posts: 200
Loc: Haaltert, Belgium
|
|
And the larger version
|
Aza
sage
Reged: 04/08/08
Posts: 203
Loc: UK
|
|
Very nice job Bart! I think thats the best version of the data I have seen
Thanks for taking the time to process it!
--------------------
Arran
My main website
My blog
|
Previous
Index
Next
Threaded
Edit
Reply
Quote
[Re: 
