Return to the Cloudy Nights Telescope Reviews home page


Astrophotography and Sketching >> CCD Imaging & Processing

Pages: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | (show all)
Inverted
sage
*****

Reged: 01/19/13

Loc: LP Land
Re: 5 or 30 minute exposures for narrowband? new [Re: freestar8n]
      #5655995 - 02/01/13 10:12 AM

Quote:

It is all about making the read noise small compared to other camera noise, because read noise is the only thing that can have reduced impact through longer sub-exposures. It also assumes a cold camera so dark current can be ignored - which may not be the case in general - and sky noise is all that matters.





That's the thing, if the models are correct and the only thing that prevents us from using more short subs, is read noise, the better calibration should eliminate much of that issue.

If we go back to the dart example, it's sort of like throwing the darts, at a bar, with your friends. Intuitively, you can get the same result whether you throw lots of darts at once, or stop and start again latter. However, read noise is like the friends you brought along. Every time you stop, they decide it will be fun to spin you around. Now, when you go back, you are dizzy, so your throws become more variable and you still feel like your spinning in one direction, so, there is a "bias" added as your throws tend to drift I one direction.

However, you may realize that your friends calm down after some number of beers, but they get more rowdy if they have too many. So, you can use the same principles used to find the bullseye, to find the optimal amount of beer, so that your friends have less of an effect. The problem is you get diminishing returns as you try to further refine your estimation of the optimal amount of beer. If you try to measure this 16 times, there is still around 25% variation in your ability to determine the correct amount of beer. If you do 20 measures, you should still have a 22% variation in your certainty of the correct amount. So, it seems like there is no real gain. However, I you went to 400 measures, your guess would now be accurate to within 5% of the optimal level. If your estimation of the bullseye location is mainly limited by your friends, you've now effectively reduced they influence by 95% compared to 75% with just 16 measures. And if the bullseye is really small, that little percentage may make a huge improvement in your ability to find it. It may still not be enough enough of a reduction though, to make shorter throwing sessions equivalent to longer ones, but it gets you closer. You can now go much shorter than before calibration and achieve the same result.

If you further and further improve your calibration of your friends (read noise) at some point, other, more nominal sources of error may become a bigger proportion of the remaining limitation for further accuracy. Perhaps now that your friends are not bothering you, the drunks at the bar are becoming more noticeable (think of them as sky noise, cosmic rays, or passing clouds, or periods of poor "seeing" etc...). So, perhaps it is beneficial to pause and deal with you now mellow friends a little more, so that you can redo these sets of throws. Or perhaps, you want to do this because you are out of shape and your muscles are getting tired (think poor polar alignment. Or equipment limitations). Then though, a narrow band filter is sort of like some headphones, you can put on some nice soothing music and forget all about the drunks and just carry on however, this doesn't help your muscles and unfortunately the headphones aren't perfect, so, at some point we may still want to retry some sets do to other although more nominal influence of the drunks....

Edited by Inverted (02/01/13 10:32 AM)


Post Extras: Print Post   Remind Me!   Notify Moderator  
Mike7Mak
Pooh-Bah
*****

Reged: 12/07/11

Loc: New York
Re: 5 or 30 minute exposures for narrowband? new [Re: Inverted]
      #5656085 - 02/01/13 10:42 AM

Whoa, I hope you're not expanding that dart analogy strictly for my benefit cuz I'm finding it harder to fathom than the technical explanation it's meant to represent.

Post Extras: Print Post   Remind Me!   Notify Moderator  
Inverted
sage
*****

Reged: 01/19/13

Loc: LP Land
Re: 5 or 30 minute exposures for narrowband? new [Re: Mike7Mak]
      #5656115 - 02/01/13 10:55 AM

No LOL. I figure, as I apparently can't summarize my thoughts do to different uses of terminology... I'd summarize with humor, so I can put this to away and go back to looking at pretty pictures

Post Extras: Print Post   Remind Me!   Notify Moderator  
Fogboundturtle
professor emeritus


Reged: 05/20/09

Loc: Burnaby, BC
Re: 5 or 30 minute exposures for narrowband? new [Re: Inverted]
      #5656126 - 02/01/13 11:00 AM

This thread is the result of too many clouds. Otherwise everyone would be busy photon capturing.

Post Extras: Print Post   Remind Me!   Notify Moderator  
korborh
professor emeritus
*****

Reged: 01/29/11

Re: 5 or 30 minute exposures for narrowband? new [Re: Fogboundturtle]
      #5656180 - 02/01/13 11:29 AM

This thread is awesome. While I do like capturing photons, a large part of what attracts me in this hobby is the opportunity of being able to experiment and understand the underlying physics conspiring against accurate observations of nature.

Post Extras: Print Post   Remind Me!   Notify Moderator  
Mike7Mak
Pooh-Bah
*****

Reged: 12/07/11

Loc: New York
Re: 5 or 30 minute exposures for narrowband? new [Re: korborh]
      #5656216 - 02/01/13 11:57 AM

There was something mentioned early on that I meant to question but forgot. And if I missed the explanation along the way my apologies for the repetition.

It had to do with longer, hence fewer, exposures "injecting a smaller multiple of the read noise into the final stack for a fixed total integration time".

That seems to imply that read noise accumulates/increases with each frame added to the stack instead of diminishing at the square root blah blah rate of other noise.

I'd chalk it up to me reading it wrong except I've read similar statements elsewhere that gave me the same impression. What am I missing?


Post Extras: Print Post   Remind Me!   Notify Moderator  
Peter in Reno
Postmaster
*****

Reged: 07/15/08

Loc: Reno, NV
Re: 5 or 30 minute exposures for narrowband? new [Re: dan17]
      #5656283 - 02/01/13 12:37 PM

Quote:

Many things will be clear and easy to understand if You all have a look at this:

http://www.cloudynights.com/item.php?item_id=1966

http://www.cloudynights.com/item.php?item_id=1973

http://www.cloudynights.com/item.php?item_id=2001

http://www.cloudynights.com/item.php?item_id=2042

http://www.cloudynights.com/item.php?item_id=2394




Craig Stark is one of the best writer in the Astronomy market. He not only writes well, but explains in layman's terms. Please read his articles.

Peter


Post Extras: Print Post   Remind Me!   Notify Moderator  
pfile
Post Laureate


Reged: 06/14/09

Re: 5 or 30 minute exposures for narrowband? new [Re: Peter in Reno]
      #5656379 - 02/01/13 01:34 PM

i think if you could calibrate away read noise then the concept of sky-limited vs. read noise limited exposures would not even exist - the idea of exposing long enough to clear the read noise would not exist.

Post Extras: Print Post   Remind Me!   Notify Moderator  
Alph
Carpal Tunnel


Reged: 11/23/06

Loc: Melmac
Re: 5 or 30 minute exposures for narrowband? new [Re: Mike7Mak]
      #5656552 - 02/01/13 03:06 PM

Quote:

That seems to imply that read noise accumulates/increases with each frame added to the stack instead of diminishing at the square root blah blah rate of other noise.




Noise (random) always adds up whether you are subtracting or adding images. There are basically two types of noise in CCD: Poisson noise and readout noise. Poisson noise is dark current shot noise, sky-glow photon noise and object photon noise. Regardless of random noise type, the noise always adds quadratically.


Post Extras: Print Post   Remind Me!   Notify Moderator  
mikeschuster
super member


Reged: 08/25/11

Loc: SF Bay area
Re: 5 or 30 minute exposures for narrowband? new [Re: Alph]
      #5656599 - 02/01/13 03:37 PM

Alph, are you implying that stacking beats down poisson noise more or at a higher rate than read noise? If so that is not what I experience, they seem to go down at the same rate. On the other hand, in my experience, FPN does not get beat down at all with stacking.
Mike

Edited by mikeschuster (02/01/13 03:40 PM)


Post Extras: Print Post   Remind Me!   Notify Moderator  
cn register 5
scholastic sledgehammer


Reged: 12/26/12

Re: 5 or 30 minute exposures for narrowband? new [Re: mikeschuster]
      #5656673 - 02/01/13 04:18 PM Attachment (9 downloads)

I spent the afternoon collecting darks, and used DSS to process the images I took the other evening. I managed to use 14 of the frames but some are pretty obscured.

I ended up with this, by APOD standards it's terrible, but it's about the best I've managed so far.
Oynx 80mm F/6.25 scope, ATK383L+ binned 2x2, 17 x 5 minutes & 16 x 5 minutes darks. Aligned and stacked with DSS. Minimal processing, a bit in DSS and a curve in PS3.

Chris


Edited by cn register 5 (02/01/13 04:29 PM)


Post Extras: Print Post   Remind Me!   Notify Moderator  
Inverted
sage
*****

Reged: 01/19/13

Loc: LP Land
Re: 5 or 30 minute exposures for narrowband? new [Re: pfile]
      #5656689 - 02/01/13 04:28 PM

Quote:

i think if you could calibrate away read noise then the concept of sky-limited vs. read noise limited exposures would not even exist - the idea of exposing long enough to clear the read noise would not exist.




Yes, I think so! Of course we can't really calibrate out read noise entirely, but I think you can do a lot more than some people do. If someone assumes that the can't do better because of diminishing returns, and stop at some number, such as 16, then our estimate of the FPN part of the read noise, still has a lot of variance ("noise"). On my camera, it is about a 8 second download, plus figure, in say idk 2 seconds between frames, that's 10 seconds per bias frame. There are 86400 second in a day, so, if I had a cloudy day to waste, I could theoretically take 8640 bias frames. If I took that many, assuming the random component was fairly normal to begin with (or at least more or less followed some exponential family PDF), then I would reduce the random component to 100*(1/sqrt(8640)) = 1.07% of the initial random noise value noise value. So, my asymptotic estimate of the FPN would be pretty good LOL. Of course, things such as "moving pattern noise" may follow very different distributions and take a lot longer to converge. So, having a chip with nice, Gaussian properties to begin with should help a lot too...


Edit: by the way, just to clarify, I'm asuming there is a higher ratio of pattern noise to random noise. I think that was obvious, it re-reading I see how it could be interpreted that I am saying we are calibrating out random noise. Of course you can't. That's why I keep saying if we have low read noise AND good calibration. As I said I sti don't have a sense o the magnitudes. That is what I'm interested in though. Hopefully once I have a chance to read Craig's article, that will be more clear.

Edited by Inverted (02/01/13 05:50 PM)


Post Extras: Print Post   Remind Me!   Notify Moderator  
Inverted
sage
*****

Reged: 01/19/13

Loc: LP Land
Re: 5 or 30 minute exposures for narrowband? new [Re: mikeschuster]
      #5656730 - 02/01/13 04:45 PM

Quote:

Alph, are you implying that stacking beats down poisson noise more or at a higher rate than read noise? If so that is not what I experience, they seem to go down at the same rate. On the other hand, in my experience, FPN does not get beat down at all with stacking.
Mike




Every-time you take an image, the signal can be thought of as having a standardized value of "1". The "random" part of the noise is just "variance" of a normal distribution. It is 1/sqrt(N) * stdev So, if we standardize the mean signal to "1" then the stdev of it is also "1". So, the noise is just 1/sqrt(n). It's my understanding from this thread that the FPN is another component of "read noise", separate form the random noise. However, this pattern is the same everytime you take an exposure, so, to make things simple, we can also standardize this to "1". So, each time you take an image, you get 1 unit of FPN from the camera read and 1 unit of signal from the DSO. The random part of the noise though is 1/sqrt(n). So, the signal and FPN accumulate faster and eventually the random noise becomes tiny in comparison.


So, as I just said above ^^ you should be able to improve your bias frames, in order to remove the FPN. You shouldn't need to take 8640 though LOL, 100 should remove 90% and should only take maybe a little over 15 minutes depending on your camera. I suspect that may help. It sounds like the bias should also be in your darks, but it takes longer to take 100 darks. Also, once you create a master bias frame, you should be able to delete the 100, so, it shouldn't kill your HD space. Of course, I'm no expert astrophotographer, so, an expert may have other suggestions..


Post Extras: Print Post   Remind Me!   Notify Moderator  
bilgebayModerator
Post Laureate
*****

Reged: 11/06/08

Loc: Türkiye - Istanbul and Marmar...
Re: 5 or 30 minute exposures for narrowband? new [Re: Inverted]
      #5656733 - 02/01/13 04:46 PM

and here is a frame with no dark calibration at all, from a different camera and scope



Hires photo

6x 1200 + 1x 600seconds Light calibrated with flat and bias master frames
FSQ106EDXIII
Astrodon 3nm Ha filter - SX Filterwheel
Atik 460 EXM
Vixen New Atlux Mount + NexAtlux board - controlled via Sky Safari
Guided via 50mm Borg + SSAG/PHD
Framed, focused and captured with Sequence Generator Pro
Calibrated, aligned and stacked CCDStack
Curves and levels in Photoshop.


Post Extras: Print Post   Remind Me!   Notify Moderator  
Alph
Carpal Tunnel


Reged: 11/23/06

Loc: Melmac
Re: 5 or 30 minute exposures for narrowband? new [Re: mikeschuster]
      #5656740 - 02/01/13 04:48 PM

Quote:

If so that is not what I experience, they seem to go down at the same rate.



I have updated my post. You are essentially right, however noise never goes down. It is the S/N that goes up when stacking.


Post Extras: Print Post   Remind Me!   Notify Moderator  
bilgebayModerator
Post Laureate
*****

Reged: 11/06/08

Loc: Türkiye - Istanbul and Marmar...
Re: 5 or 30 minute exposures for narrowband? new [Re: bilgebay]
      #5656750 - 02/01/13 04:51 PM

a DSLR frame



Hires photo

This one was shot under moon light and terrible seeing conditions.

6x 1200 seconds Light @ ISO 1600, dark, flat, and bias master frames
AT110EDQ+AT Field Flattener
Astronomik 6nm Ha filter
Canon T2i, Hutech modified
Vixen New Atlux Mount + NexAtlux board - controlled via Sky Safari
Guided via Lacerta OAG + SSAG/PHD
Framed, focused and captured with BackyardEOS
Calibrated, aligned and stacked with Images Plus
Curves and levels in Photoshop.

File Name HORSEHEAD_LIGHT_1200_1600iso_HA_6NM_+25c_12c_AT110+FF_AMBIENT12c_777_1829_20111106_05h39m05s825_000429.CR2
Camera Model Canon EOS REBEL T2i
Firmware Firmware Version 1.0.8
Shooting Date/Time 06.11.2011 06:19:44
Author BILGEBAY
Owner's Name
Shooting Mode Manual Exposure
Tv( Shutter Speed ) 1197
Av( Aperture Value ) 0.0
Metering Mode Evaluative Metering
ISO Speed 1600
Auto ISO Speed OFF
Image Size 5184x3456
Image Quality RAW
Flash Off
FE lock OFF
White Balance Mode Custom
AF Mode Manual focusing
Picture Style Faithful
Sharpness 0
Contrast 0
Saturation 0
Color tone 0
Color Space Adobe RGB
Long exposure noise reduction 0:Off
High ISO speed noise reduction 3:Disable
Highlight tone priority 0:Disable
Auto Lighting Optimizer 3:Disable
Peripheral illumination correction Disable
File Size 29342 KB
Dust Delete Data No
Drive Mode Single shooting


Post Extras: Print Post   Remind Me!   Notify Moderator  
bilgebayModerator
Post Laureate
*****

Reged: 11/06/08

Loc: Türkiye - Istanbul and Marmar...
Re: 5 or 30 minute exposures for narrowband? new [Re: bilgebay]
      #5656770 - 02/01/13 05:00 PM

One last frame from yet another camera, QSI683.... Luminance only, properly calibrated with bias, dark and flat frames.

13x 600 seconds



Hires


Post Extras: Print Post   Remind Me!   Notify Moderator  
pfile
Post Laureate


Reged: 06/14/09

Re: 5 or 30 minute exposures for narrowband? new [Re: Inverted]
      #5656878 - 02/01/13 06:02 PM

Quote:


Edit: by the way, just to clarify, I'm asuming there is a higher ratio of pattern noise to random noise. I think that was obvious, it re-reading I see how it could be interpreted that I am saying we are calibrating out random noise. Of course you can't. That's why I keep saying if we have low read noise AND good calibration. As I said I sti don't have a sense o the magnitudes. That is what I'm interested in though. Hopefully once I have a chance to read Craig's article, that will be more clear.




here is how you quantify the read noise.

make a really good master bias frame.
subtract the master bias frame from a bias sub.
the result is the read noise in that bias sub.


Post Extras: Print Post   Remind Me!   Notify Moderator  
Mike7Mak
Pooh-Bah
*****

Reged: 12/07/11

Loc: New York
Re: 5 or 30 minute exposures for narrowband? new [Re: Alph]
      #5656900 - 02/01/13 06:15 PM

Quote:

...however noise never goes down. It is the S/N that goes up when stacking.



Doh, ok that's the bit I was missing.

So since the corresponding signal (bias/pedestal) that created the read noise is subtracted out the noise is left behind to accumulate. Which is why covering it with sky noise is the only way to eliminate its effect. Is that close?

Thanks.

Edited by Mike7Mak (02/01/13 06:17 PM)


Post Extras: Print Post   Remind Me!   Notify Moderator  
Alph
Carpal Tunnel


Reged: 11/23/06

Loc: Melmac
Re: 5 or 30 minute exposures for narrowband? new [Re: Mike7Mak]
      #5656964 - 02/01/13 06:59 PM

Quote:

So since the corresponding signal (bias/pedestal) that created the read noise is subtracted out the noise is left behind to accumulate. Which is why covering it with sky noise is the only way to eliminate its effect. Is that close?




Yes, you are pretty close.
The random residue of readout noise and/or dark current noise is always left behind and covering it with the sky glow or a nebula is the only way to hide it from our eyes


Post Extras: Print Post   Remind Me!   Notify Moderator  
Pages: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | (show all)


Extra information
17 registered and 22 anonymous users are browsing this forum.

Moderator:  Oldfield, Scott in NC, bilgebay, WOBentley 

Print Thread

Forum Permissions
      You cannot start new topics
      You cannot reply to topics
      HTML is disabled
      UBBCode is enabled


Thread views: 7391

Jump to

CN Forums Home




Cloudy Nights Sponsor: Astronomics