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]
      #5659545 - 02/03/13 09:05 AM

Quote:


As for lower read noise - as I've been saying - it all depends on your camera, optics, and site. You can turn a "bright" site into a dark one just by increasing the f/ratio or using a narrower filter. It's all about the actual sky flux you are receiving and the actual read noise you have.





Yes, those parameters are interesting to visualize too. That makes sense from the posted models. I was tempted to try to plug in the models and graph the numbers myself, but didn't get around to it yet.

I don't mean to give Frank homework either by the way, but if he happens to play around with the numbers and sees anything interesting let us know! Or let me know what model you are using and I may play around with it.


Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: 5 or 30 minute exposures for narrowband? new [Re: Inverted]
      #5659800 - 02/03/13 11:34 AM Attachment (14 downloads)

Well for now I'll provide my own worked example using a Pacman Ha imaging session. This is C11 at f/5.7 on cge-pro with OAG and, of course, MetaGuide in 15m exposures with 2.0" fwhm. SXVF-H9 Sony ccd with measured values of gain = 0.4 e/adu and read noise of 13.5e. It is supposed to be 12e, but unfortunately is 13.5e. I measured a sky flux of 0.18e/s from a calibrated sub, and bias frames have a value around 750 adu, so it is a huge offset that must be removed. The Ha filter is Baader 7nm and I have a fair amount of light pollution - so I have neither a low read noise camera, nor a dark sky with narrow filter.

The question is - is 15m long enough or would I get a big win going much longer? I can guide longer, but I'm mainly concerned about losing subs at 30m - which would defeat any win I would get from reducing read noise. So here is the comparison between 15m subs and 30m subs for a session of about 3 hours:


Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: 5 or 30 minute exposures for narrowband? new [Re: freestar8n]
      #5659813 - 02/03/13 11:40 AM Attachment (12 downloads)

It's clear that I'm hurting from both sky glow and read noise - but there isn't much win going from 15m subs to 30m, and in fact with only 3 hours allocated I would have many fewer subs to average. If I used a narrower filter that would help - but then read noise would really dominate. So - as always - it's good to have dark site, low read noise, and narrow filter - but it's also good to make use of what you have in an optimal way for the given conditions - and 15m exposures worked ok here. Here is a close up of a simple average of 11 subs with some DDP stretch. It has some noise in the darkest regions, but not too bad. It is at 50% size - the original being 0.83"/pixel.

Frank


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: freestar8n]
      #5659986 - 02/03/13 01:20 PM

Bottom line is NB wins for bright site. Look at my Bubble Nebula in "Peter's Galleries" in my signature. Ha/Oiii were 15 minutes subs and Sii was 30 minutes subs. My Atik 460EX camera has very low read noise and it really helps. My web site shows Hubble Palette, Bi-color Ha/Oiii/Oiii, HaRGB, RGB using pseudo Luminance extracted from RGB. I could not use Luminance because even at 10 minutes sub-exposures, the stars were badly bloated and the sky background was too bright. The histogram for Luminance showed the left side of graph shifted too far to the right. My calibrated processes use Bad Pixel Mapping instead of dark subtraction. I get better results with BPM.

When I imaged with Sii filter, I started at 15 minutes. I examined the first image downloaded from my camera and decided to image at 30 minutes subs for the rest of the night. The calibrated 30 minute sub appear to look better than calibrated 15 minute sub.

I uploaded calibrated Sii with BPM, flat and bias of 15 and 30 minutes subs for you to evaluate at:

https://www.dropbox.com/s/ot1ftleoov4cale/Bubble_SII_15minutes_calibrated.fit

https://www.dropbox.com/s/7idi0rwnnm690nl/Bubble_SII_30minutes_calibrated.fit

Note that each image is not stacked, just one calibrated sub of each. I don't have fancy tools like Frank does to do measurements so when you have time, please look at my sample subs and report back if you can. I appreciate it. I think it was worthwhile imaging at 30 minutes over 15 minutes. Fortunately not one single airplane or satellite went by all night! I live directly under the airplanes' flight paths. Yes, my house vibrates when planes fly over.

This is probably the best thread I've read in a long time. Thanks to OP for starting this thread. Also, it turned out that I really screwed up the calibration of Sii for 30 minute subs of Bubble Nebula when I was processing Hubble Palette. Now I have to re-process it all over again but that's okay and best of all it's fun.

Thanks,
Peter


Post Extras: Print Post   Remind Me!   Notify Moderator  
korborh
scholastic sledgehammer
*****

Reged: 01/29/11

Loc: Arizona
Re: 5 or 30 minute exposures for narrowband? new [Re: Peter in Reno]
      #5660040 - 02/03/13 01:57 PM

Frank, great discussion and thanks for taking the effort to plot graphs and explain the various scenarios. It has really given a crisp and clear picture of the trade-offs based on equipment/conditions.

I will doing some more analysis on my recent narrowband images. I am currently going 30min subs with 5nm Ha filter from bright sky and low read noise (~3.5e-) camera. Seems like I may be better off with something like 20mins with less risk of losing subs to focus-shift and UFO's. Need to try in the next imaging session.


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: cn register 5]
      #5660175 - 02/03/13 03:12 PM

Quote:

The dark sky results show that increasing exposure beyond the 5 minutes of your example would be useful, and this explains why the experts at dark sky sites insist that longer exposures are beneficial, they are much more so for them.

I wonder, how long professional astronomers expose for, they have perfect equipment at very dark sites and are trying to get the most out of very dim objects. Does anyone know how long the individual exposures in the Hubble deep field were?




First of all, professional astronomers are not interested in pretty pictures. They have no scientific value.
Secondly, professional astronomers are not concerned how the image background looks like. All what they care is the S/N of the object of interest. Somehow the object S/N is completely missing from the Frank’s presentation. I am afraid this will only add to the misconception that you need sky glow in your images.


Quote:

As for the claim that increasing exposure length, even from from 10 to 15 minutes, makes a difference to resolution and contrast we have been given no evidence that this is so




There is nothing fundamentally wrong with that statement. A higher S/N results in a higher contrast that can be interpreted as a higher spatial resolution.
As a related matter that has not been mentioned in this thread. A brighter sky requires more total exposure time than a dark sky to achieve the similar contrast.


Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: 5 or 30 minute exposures for narrowband? new [Re: Peter in Reno]
      #5660345 - 02/03/13 04:56 PM

Hi Peter-

I'd be happy to look at your numbers - but I do need gain and read noise for your camera. If you can measure them that would be great.

I didn't mention it - but I don't use dark subtraction either with my Sony ccd. If the dark current is low and there is no pattern noise in the dark except for hot pixels - then it's just a hot pixel map and it doesn't help to subtract.

Thanks,
Frank


Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: 5 or 30 minute exposures for narrowband? new [Re: korborh]
      #5660355 - 02/03/13 05:00 PM

Thanks Korhorh-

I have appreciated your positive comments in this thread - and those from others. 3.5e sounds great - and would nicely complement the rather high noise data I posted here.

Thanks,
Frank


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: freestar8n]
      #5660492 - 02/03/13 06:05 PM

Hi Frank,

Try these numbers I found from another person's Atik 460EX camera:

Gain = 0.26
Readout Noise = 4.79e-
Total System Noise = 5.10e-

Temperature was at -15C. My images were at 0C. It's my understanding that temperature colder than 0C makes small difference for Sony CCD cameras as compared to Kodak CCDs.

Atik 460EX Factory spec:
Read Noise: 5e-

Atik does not specify the gain for Atik 460EX. SX H694 specify the System Gain of 0.3 electrons per ADU.

My Bias frames have minimum of 300 ADU, Mean of 335 ADU and maximum of 365 ADU.

Thanks,
Peter

Edited by Peter in Reno (02/03/13 06:09 PM)


Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: 5 or 30 minute exposures for narrowband? new [Re: Peter in Reno]
      #5661128 - 02/04/13 04:27 AM Attachment (6 downloads)

Hi Peter-

It looks like your 15m has a background level of about 70 adu and 30m has 140 adu so the sky background ends up at 0.02e/s using gain 0.26. With read noise of 4.8e I generated the plots below. You have a much darker background but your read noise is also much lower than mine - so 15m is pretty good. 30m helps a bit, but still doesn't get you down to the actual sky background.

Regarding SNR - the noise models I'm referring to are exactly that - a calculation of the fundamental background noise in an image. This is important both for aesthetic imaging, since it determines how noisy the dark regions look, and for scientific work, since it places a limit on the SNR of a given object. The background noise is well defined, but the SNR depends entirely on which signal you are interested in. In a given image, the SNR will vary from 0 in an area of pure background, to perhaps hundreds in a bright patch of nebulosity.

The other complication is that the nebulosity signal itself is Poisson, so it has intrinsic shot noise that affects the SNR. So - a faint signal will have SNR limited by the background and:

SNR(faint) = SFaint/NBkg

an intermediate signal will have a mixture:

SNR(intermediate) = SMed/sqrt(NBkg^2+SMed^2)

and a very strong signal will just be limited by its own shot noise:

SNR(High) = sqrt(SHigh)

- where all these values are electrons - not adu.

Frank


Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: 5 or 30 minute exposures for narrowband? new [Re: freestar8n]
      #5661129 - 02/04/13 04:32 AM

Pater -one other note: you should probably study the actual dark current you are getting at that temperature - by comparing 15m darks and 30m darks. If there is, in fact, some dark current causing the background signal, then you might have a way to reduce it - as opposed to sky background that you are stuck with.

Frank


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: freestar8n]
      #5661258 - 02/04/13 08:11 AM

Neat, so, it looks like the ratio between the dark 15 and 30m subs is the same in both cases. However, the magnatude of the effect is almost 3.5x larger in the higer read noise case. That would seem to make sense, now if we could figure out exactly how much of an effect can become apparent after proccessing and all that good stuff

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: Alph]
      #5661290 - 02/04/13 08:44 AM

Quote:

Somehow the object S/N is completely missing from the Frank’s presentation. I am afraid this will only add to the misconception that you need sky glow in your images.





True, I think. Really it seems it would make more sense to normalize and then subtract the total noise from 1.

Of course, it's also worth mentioning all this seems to assume we've met nyquest criteria for sub exposure times (adjusting for quantization error) and n number of sub exposures is a reasonable number (i.e. > 10ish).


Post Extras: Print Post   Remind Me!   Notify Moderator  
freestar8n
Post Laureate
*****

Reged: 10/12/07

Re: 5 or 30 minute exposures for narrowband? new [Re: Inverted]
      #5661314 - 02/04/13 09:02 AM

The goal here is to provide a number that conveys the relative impact of read noise in a given imaging situation - and the potential benefit of longer exposures. It does this by expressing the net impact of read noise relative to sky noise. I think it serves the purpose well - but will still require some level of understanding by anyone who might use it, along with a sense of proportionality.

The combination of the numbers with the plots above will, I hope, make the whole thing even clearer.

I think there is already evidence that based on people reading about "optimum sub exposures" and various write ups on the web, some concluded that images would be noisier at a dark site because the darkness lets the noise "show" more. That is incorrect, and direct plots of the accumulated noise in different scenarios should show that clearly.

Frank


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


Reged: 12/09/09

Loc: Merced CA
Re: 5 or 30 minute exposures for narrowband? new [Re: freestar8n]
      #5661456 - 02/04/13 10:30 AM

You tell that to John Smith, Frank. I'd bet he'll disagree with you. You can email him at john@ccdware.com It's not that the same noise isn't there at a dark site, it just shows more because the sky glow is signal and that "signal" hides the noise at a more LP site. Of course that sky glow signal has other problems with it such as limiting exposures.

Post Extras: Print Post   Remind Me!   Notify Moderator  
shams42
Pooh-Bah
*****

Reged: 01/05/09

Loc: Kingsport, TN
Re: 5 or 30 minute exposures for narrowband? new [Re: vpcirc]
      #5661467 - 02/04/13 10:37 AM

The reason skyglow is undesirable is because it carries Poisson shot noise equal to the square root of the skyglow signal. Otherwise it could be dealt with by a levels adjustment.

If your sub has 1000e- skyglow, it will have 100e- noise due to skyglow. If you go to a dark site and now get 100e- skyglow in your sub, you will only have 10e- noise from skyglow.

Skyglow adds noise. Why else would one seek out a dark site?


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


Reged: 12/09/09

Loc: Merced CA
Re: 5 or 30 minute exposures for narrowband? new [Re: shams42]
      #5661505 - 02/04/13 10:49 AM

Dark sites allow much more of the signal from the DSO to reach the chip as the natural light is not being "polluted" by man made light sources. Contrast and resolution improve dramatically. But I'm done making my point. Frank is the expert here who is never wrong. You're right Frank and all the information I've been given is totally false.

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: freestar8n]
      #5661578 - 02/04/13 11:30 AM

Quote:

Pater -one other note: you should probably study the actual dark current you are getting at that temperature - by comparing 15m darks and 30m darks. If there is, in fact, some dark current causing the background signal, then you might have a way to reduce it - as opposed to sky background that you are stuck with.

Frank




Hi Frank,

Thanks for the report.

I uploaded a Master Dark of 15 minutes (stack of 8 subs) and 30 minutes (stack of 6) if you are interested. I didn't need a lot of dark subs because Bad Pixel Mapping does not require many dark subs. Both of them at 0C.

https://www.dropbox.com/s/1wk8pyx9961qksj/Atik460EX_Master_Dark_15min.fit

https://www.dropbox.com/s/us8vna4npqdn2nq/Atik460EX_Master_Dark_30min_0C.fit

I also uploaded a Master bias of 30 subs at 0C:

https://www.dropbox.com/s/qqnxbg56i3me2k0/Atik460EX_Master_Bias.fit

Peter


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

Reged: 11/06/08

Loc: Turkiye - Istanbul and Marmari...
Re: 5 or 30 minute exposures for narrowband? new [Re: Peter in Reno]
      #5661589 - 02/04/13 11:37 AM

Quote:

Hi Frank,

Try these numbers I found from another person's Atik 460EX camera:

Gain = 0.26
Readout Noise = 4.79e-
Total System Noise = 5.10e-

Temperature was at -15C.

Thanks,
Peter




Hi Peter,

Do these figures belong to my 460EX?


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: vpcirc]
      #5661593 - 02/04/13 11:38 AM

Quote:

You tell that to John Smith, Frank. I'd bet he'll disagree with you. You can email him at john@ccdware.com It's not that the same noise isn't there at a dark site, it just shows more because the sky glow is signal and that "signal" hides the noise at a more LP site. Of course that sky glow signal has other problems with it such as limiting exposures.





Sky-glow should converge to a fixed mean. In statistical terms, this "biases" the results (as described previously). In other words, a bright site, may have more "bias" (or "unwanted signal" if you prefer), so the "observed" signal to noise ratio may appear higher, due to this bias.

However, this is because, as mentioned a few pages back, we are only estimating the underlying signal to noise ratio, based on the observed data, which includes the bias (sky-glow). So, this estimate includes some "unwanted signal" from the sky-glow, and the variance added by the sky glow is proportionally less than the signal. So, SNR being the (signal mean) / sqrt(variance) may look higher if estimated from the observed data. However, really, if you were to properly adjust for the bias, it actually isn't. So, this is a very misleading result.

In some ways, it may make more sense, as Alph alluded to, to use to the contrast ratio. The contrast will be higher given the same amount of exposure time, from a dark site, than from a bright site. Also an unbiased SNR estimate would be higher given the same amount of exposure time, from a dark site, than from a light site. A biased SNR estimate may show that you "need" to go longer from a dark-site, to get the same SNR, as from a bright site, but this is about as mathematically relevant as showing 0 equals 1 and likewise, the solution relies on faulty logic.


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 19 anonymous users are browsing this forum.

Moderator:  Oldfield, droid, bilgebay, WOBentley, dr.who 

Print Thread

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


Thread views: 7614

Jump to

CN Forums Home




Cloudy Nights Sponsor: Astronomics