Click here if you are having trouble logging into the forums
Privacy Policy |
Please read our Terms
of Service | Signup and
Troubleshooting FAQ | Problems? PM a Red or a Green Gu.... uh, User
Kolenka
Pooh-Bah
Reged: 06/01/08
Posts: 1009
Loc: Seattle Area, WA, USA
|
|
This is a continuation of a side-topic started in this thread...
Quote:
Let us take a situation.
4" f/4 scope vs 8" f/8 scope. The 8" has 4x the light grasp or gathers 4x more light than the 4", but the image (object) is 16x larger. Because the object is 16x larger in area, it takes 16x more photons to bring the exposure to the same brightness on the chip, because the light is distributed over the area that is 16x greater. So, even though the 8" has 4x the light gathering, the exposure still requires 4x longer to achieve the same exposure.
Now, if the scope is an 8" f/4 vs 8" f/8, the image is 4x larger in area, so it will require 4x longer to expose the object due to the larger area of the f/8.
Tell me what is incorrect about this statement?
Blueman
The problem here with the statement is that you are discussing brightness in the exposure. But when dealing with AP and a CCD, we actually don't care about the real brightness of the exposure. We stretch it anyways. What we care about is the SNR and the dynamic range.
While brightness scaling is linear, SNR isn't. Not once you get the signal well above the read noise. Once you hit a certain exposure time, the characteristics of the camera, the subject, and the skyglow take over and determine your SNR. Until you reach that point, you will see a fairly linear effect on SNR. But assuming you are above the read noise, SNR is not linear with exposure time, unlike brightness.
EDIT: Craig's latest article here on Cloudy Nights discussing SNR provides a pretty good example that SNR and brightness are actually different beasts.
--------------------
Orion XX12 / Orion 80ED OTA / AT66ED
Nagler 7T6, 9T6, 13T6, 17T4, 26T5
Canon XS, TIS DMK 31AF03, AstroTrac TT320X
Northwest Astro Photoblog
Edited by Kolenka (05/05/09 03:25 AM)
|
tiggere
super member
Reged: 02/22/08
Posts: 145
|
|
I guess I'm guilty of the same thinking as blueman...I was under the assumption that from F4 to F8 is two full F-stops so it would take four times as long...as F8 is only seeing 1/4 of the light that F4 is...
The other issue is exactly how much does aperture play a role...lets use a real case scenario...lets assume Mag 7 skies and shooting M51 (I'm gonna get a good shot of that darn thing yet !!)...I am using a 6" SCT @ F6.3 (with the reducer)with 28.27 sq. inches of mirror and was wanting to move to a 8" RC @ F6 (with the reducer) with 50.27 sq. inches of mirror...the F-stops are about equal or close enough for government work anyhow...focal lengths will also be within a 100mm...so will it still take the same time (minutes)for the shot or will the aperture speed things along...and how much...
Now we can compare the 8" AT8IN Newt @ F4 and 800mm FL to the 8" RC at F6 and 1016 (or there abouts) FL...aperture is the same and its just over one F-stop by a hair...not sure the focal length will make a difference...so comparing these two how much longer(in minutes)would it take to get the same shot of M51 under Mag 7 skies...
Lets also use minutes of the sub for the answer instead of 1/3 as long as we are all looking to save time and everything we do is based on that anyhow...
--------------------
Celestron C6A-XLT
Atlas EQ_G with EQMOD / HNSKY
Logitech Dual Action Controller
Philips SPC900NC Webcam
Wilmington, NC
|
Miguel Lopes
professor emeritus
Reged: 01/04/07
Posts: 694
Loc: Portugal
|
|
tiggere, you've left out the most important thing: the CCD!
That is why the term "fast" is not accurate for CCD's, especially those where you can bin the pixels!
I think the f ratio is important for image scale, optical defects/tolerances and backfocus. Everything else does not apply to the CCD world.
--------------------
Astrology is the science for ignorants. Astronomy is the science for those who feel ignorant. - Miguel Lopes
|
Nocturnal
Pooh-Bah
Reged: 09/14/05
Posts: 1025
Loc: CT, USA
|
|
Comparing scopes with different focal lengths is difficult and gets misused in some cases. Look at advertisements for the hyperstar for example. It says exposure lengths are vastly reduced at F/2 compared to native F/10. The image scale is completely different as well so it means, IMO, nothing. To push it into the ridiculous you could cram all the light of M42 into 1 pixel and saturate it in a fraction of a second with the right optics. But you can't see any detail. Pointless.
My take on the whole thing is that you decide what image scale you want and then get the largest clear aperture you can afford or carry on your mount. Image scale depends on focal length and pixel size. Clear aperture depends on actual aperture and the central obstruction.
Twice the clear aperture (surface wise) at the same focal length will deliver twice the photons at the same image scale. *That* you can compare. *That's* why you don't see 66mm refractors mounted in those big Keck domes 
DSLRs make things more complicated because of the ISO settings. Still the physics (photons per pixel) are the same. How they translate into ADUs and noise gives me a headache so I won't attempt to understand it. CCDs are easy comparatively.
--------------------
Best,
Sander
C11-HyperStar on Atlas EQ-Q driven by EQMOD
William Optics M110 With FR-III/TRF-2008
DS2090 guide scope
QHY-8, DSI-Pro and DSI cameras
watec 802h video camera with KIWI OSD
Astro stuff: http://www.tungstentech.com
My Astro Photos
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Consider two telescopes with the same aperture but different focal lengths - an 8" f/5 and an 8" f/10. And consider photographing the same extended object with each, for the same exposure time and with the same camera.
In each case the number of photons reaching the sensor from the extended object is statistically the same, because that depends only on the aperture and exposure time. But with the f/5 telescope those photons cover one quarter of the area (number of pixels) on the sensor, so each pixel covered by the object receives four times as many photons from the object, which means four times the illumination, and thus four times the signal.
Once you get past the read noise, the primary contributor to noise in DSLR and CCD photography is shot noise, which increases proportional to the square root of illumination (NOT exposure time). So the image of the extended object captured by the f/5 telescope has four times the signal and twice the shot noise, compared to the image captured by the f/10 telescope. Therefore the SNR due to shot noise is twice as high in the f/5 telescope compared to the f/10 telescope.
The portion of the image containing the extended object captured by the f/10 telescope has more resolution (four times as many pixels covering the object), but only one quarter of the signal per pixel and one half of the SNR contributed by the major noise source, shot noise.
--------------------
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 (05/05/09 09:36 AM)
|
tiggere
super member
Reged: 02/22/08
Posts: 145
|
|
Sorry for not mentioning using a DSLR...I followed the link and didn't realise it wasn't in the DSLR forum...
So what we are doing is trading SNR for resolution by going to the higher F ratios with the downside of needing to increasing our shot times with the F10...and by increasing our shot times we are still obtaining our critical SNR numbers...
Here's what I think I'm wanting to know...if we take the 8" F/5 and 8" F/10 examples above using a DSLR with both set at ISO 800 shooting the exact same target under the exact same skies of Mag 7 on the exact same night...how much difference is there going to be on the sub lengths that will need to be shot to get the exact same results...
If the resolution is higher in the F/10 and you have to shoot 8 minute subs to to get to the optimal SNR and you only have to shoot 3 minute subs using the F/5 but have to shoot more of them to get to the same resolution as the F/10 then its 6 of one 1/2 dozen of another...the only advantage is the faster scope won't show guide errors as quick...
Is this correct?
--------------------
Celestron C6A-XLT
Atlas EQ_G with EQMOD / HNSKY
Logitech Dual Action Controller
Philips SPC900NC Webcam
Wilmington, NC
|
Kolenka
Pooh-Bah
Reged: 06/01/08
Posts: 1009
Loc: Seattle Area, WA, USA
|
|
Quote:
tiggere, you've left out the most important thing: the CCD!
That is why the term "fast" is not accurate for CCD's, especially those where you can bin the pixels!
I think the f ratio is important for image scale, optical defects/tolerances and backfocus. Everything else does not apply to the CCD world.
But that isn't entirely accurate either, and that was what spawned this whole discussion.
Say you have a graph of SNR versus exposure time. You would actually be able to split this graph into two parts, because SNR behaves differently in two regions of the graph. As you extend your time from 0 seconds, the noise is initially dominated by read noise, the slope of the line in this region is primarily determined by the f/ratio, or speed of the scope.
Once most of your signal is well above the read noise such that skyglow is now the dominant noise, you hit an inflection point of sorts, and the characteristics of the camera take over. The curve changes dramatically.
The only difference between an f/4 and f/8 scope of the same design and aperture, is that the f/4 will hit this inflection point sooner (closer to 0 seconds on the graph). Binning on an f/8 helps you reach this inflection point faster as well, but again, the camera takes over once you hit this inflection point.
Nocturnal, DSLRs are a pain to figure out here, partly because of their ISO and their noise. Because the dark current is so variable, DSLRs actually tend to push the inflection point out further compared to a cooled CCD, and will seem to support the idea that f-ratios matter when used. This is what I even encountered myself. Faster f-ratios gave me better DSLR shots, mostly because it was hard to get the signal well above the dark noise when shooting anything but open clusters.
--------------------
Orion XX12 / Orion 80ED OTA / AT66ED
Nagler 7T6, 9T6, 13T6, 17T4, 26T5
Canon XS, TIS DMK 31AF03, AstroTrac TT320X
Northwest Astro Photoblog
|
lineman_16735
Tak-o-holic
Reged: 12/04/04
Posts: 2604
Loc: Central PA
|
|
Can we at least finally agree that F ratios with CCD imaging are not the same as F ratios with film? The film F stop world just isn't the same as CCD's. An 8" F/8 to F/4 with a ccd is not going to provide you with a 4x better SNR with equal exposure times.
--------------------
Chris
A mount from Illinois
A scope from Japan
A camera from Cal-I-Fornia
A dog from Kentucky
A wife and kids from the "Twilight Zone"
The Geek Shed
|
lineman_16735
Tak-o-holic
Reged: 12/04/04
Posts: 2604
Loc: Central PA
|
|
Quote:
Now, if the scope is an 8" f/4 vs 8" f/8, the image is 4x larger in area, so it will require 4x longer to expose the object due to the larger area of the f/8.
Tell me what is incorrect about this statement?
Blueman
I really don't understand what you are saying here Floyd?
--------------------
Chris
A mount from Illinois
A scope from Japan
A camera from Cal-I-Fornia
A dog from Kentucky
A wife and kids from the "Twilight Zone"
The Geek Shed
|
WarrenS
scholastic sledgehammer
Reged: 03/04/08
Posts: 892
Loc: Orange County New York
|
|
I'm with you here on this one Chris. Having shot 35mm film daytime, for over 30 years, an f stop was an f stop, whether you used a 35mm f/2.8 or a 400mm f2.8 lens, at the same f stop the same amount of light hit the film (same iso & shutter of course). One F stop smaller was Half the light (photons or whatever) reaching the film. When I tried film A/P years back, I read that aperture has a definite effect on point light sources, ie stars, but F/Stop is what determined the amount of light hitting from an extended object, ie nebulae and galaxies.
Maybe it's different with CCD's, but until I hear it from an optical physicist or someone like that I'm skeptical.
--------------------
Warren
Astro-Tech 127EDT
Celestron Onyx 80ED
Astro-Tech Field Flattener
C8 (circa 1983 Orange Tube)
Atlas EQ-G, Orion SSAG
Canon 135mm F2.8
Canon 40D, Astronomik CLS clip filter
Leica, Minolta binos
|
Kolenka
Pooh-Bah
Reged: 06/01/08
Posts: 1009
Loc: Seattle Area, WA, USA
|
|
Quote:
I'm with you here on this one Chris. Having shot 35mm film daytime, for over 30 years, an f stop was an f stop, whether you used a 35mm f/2.8 or a 400mm f2.8 lens, at the same f stop the same amount of light hit the film (same iso & shutter of course). One F stop smaller was Half the light (photons or whatever) reaching the film. When I tried film A/P years back, I read that aperture has a definite effect on point light sources, ie stars, but F/Stop is what determined the amount of light hitting from an extended object, ie nebulae and galaxies.
Maybe it's different with CCD's, but until I hear it from an optical physicist or someone like that I'm skeptical.
The difference is that brightness is considered fixed in film, so you have a sweet spot you need to hit. With CCDs, we can stretch the brightness to fit a range of our choosing, so that is no longer an issue. But because we can stretch it, the SNR and dynamic range become more important, as they now determine how much final noise is left after we've stretched the image, and how much detail we've actually captured.
The response on a CCD also tends to be more linear than film, so exposing longer just increases the brightness after a certain point, and has a non-linear effect on SNR.
Quote:
Can we at least finally agree that F ratios with CCD imaging are not the same as F ratios with film? The film F stop world just isn't the same as CCD's. An 8" F/8 to F/4 with a ccd is not going to provide you with a 4x better SNR with equal exposure times.
I'll agree, I just don't want us to forget the caveat region where the f/4 *will* provide 4x better SNR with equal exposure times (granted we are talking about noisy cameras and/or short exposures for this to be true).
--------------------
Orion XX12 / Orion 80ED OTA / AT66ED
Nagler 7T6, 9T6, 13T6, 17T4, 26T5
Canon XS, TIS DMK 31AF03, AstroTrac TT320X
Northwest Astro Photoblog
Edited by Kolenka (05/05/09 12:33 PM)
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Can we at least finally agree that F ratios with CCD imaging are not the same as F ratios with film? The film F stop world just isn't the same as CCD's. An 8" F/8 to F/4 with a ccd is not going to provide you with a 4x better SNR with equal exposure times.
With CCD's and DSLR's, the SNR due to shot noise will be about 2X better with the faster telescope and equal exposure times - 4X more signal, 2X more shot noise.
Film is an entirely different beast once you get past a few seconds of exposure time, due to reciprocity failure. In effect the ratio of signal to exposure time with film is no longer 1:1 because the film sensitivity decreases with exposure.
--------------------
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 (05/05/09 01:28 PM)
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Once most of your signal is well above the read noise such that skyglow is now the dominant noise, you hit an inflection point of sorts, and the characteristics of the camera take over. The curve changes dramatically.
Actually sky glow isn't noise, it's signal - but it contains noise (mostly shot noise in longer exposures) and since it is the least exposed part of the frame, the SNR of the sky glow due to shot noise is the poorest SNR in the image.
Once past the read noise, the shot noise becomes the dominant noise and the SNR of all parts of the image (skyglow and subject) continues to increase with exposure time, until the entire sensor is saturated with signal. When the entire frame is white due to every pixel being completely saturated, the SNR is infinitely high, since the entire image is signal without any noise.
--------------------
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
|
Nocturnal
Pooh-Bah
Reged: 09/14/05
Posts: 1025
Loc: CT, USA
|
|
Quote:
Can we at least finally agree that F ratios with CCD imaging are not the same as F ratios with film? The film F stop world just isn't the same as CCD's.
Sorry, I can not agree with that. F ratios are a property of the OTA and have nothing to do with the recording medium. F ratio is focal length divided by aperture.
I think the confusion comes from the adjustable iris in most cameras which yields variable F ratios at a fixed focal length. Add to that adjustable focal length and it gets more complicated.
Luckily we don't deal with that with astronomical imaging. Except in rare cases aperture is set. You can fiddle with focal length with reducers and barlows but that's it. Attaching a different camera does not change the F ratio.
I don't think it makes much sense to talk F stops at all in astronomical imaging as there is no iris but maybe someone can enlighten me.
I think in this case I benefit from having very little traditional photography baggage
--------------------
Best,
Sander
C11-HyperStar on Atlas EQ-Q driven by EQMOD
William Optics M110 With FR-III/TRF-2008
DS2090 guide scope
QHY-8, DSI-Pro and DSI cameras
watec 802h video camera with KIWI OSD
Astro stuff: http://www.tungstentech.com
My Astro Photos
|
gavinm
scholastic sledgehammer
Reged: 08/26/05
Posts: 804
Loc: Auckland New Zealand
|
|
Quote:
... Maybe it's different with CCD's, but until I hear it from an optical physicist or someone like that I'm skeptical...
You may well be hearing it from physicists, but some other people still think they know better.. 
Skeptical is good BTW. Good not to believe everything you read
--------------------
Gavin
Mt Albert Grammar School Observatory
Auckland, New Zealand
http://www.mags.school.nz/astronomy/index.html
12" LX200R F6.8 AP
SBIG ST7-XME + CFW10
Moonlite SCT focuser w/ temp
Skywatcher Equinox ED80 Pro (ADM dovetail)
+ other stuff
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
I'm not sure what the points being made here are...
In order to get the same SNR as a 5 minute exposure at f/5 using an f/10 scope, you'd have to shoot for 20 minutes (4x the exposure). I know it won't be exactly the same because read noise is static and dark current noise is dependent only on exposure time and not aperture, but neither of these noise sources should be significant in a good camera at these exposure lengths.
--------------------
-Arkalius
11" Celestron SCT on Orion Atlas EQ-G
Celestron 100ED Refractor
8" Zhumell Dobsonian Reflector
|
bbasiaga
professor emeritus
Reged: 05/10/06
Posts: 724
|
|
F stops are still the same. The same histogram with my 11" SCT and hyperstar at F2 an 1 second exposure is achieved in roughly 10s on my F7 refractor. The SCT is about 3 stops faster and about 3 squared times (9) the light gathering leads to about 1/9th the exposure. It so happens the image scale on these two scopes is very similar because my 3.5" F7 refactor is only about 100mm longer (im rounding a bit here) than my SCT with hyperstar. The difference apeture makes is how we get a given focal length at a given F ratio.
The CCD, DSLR, stretching, etc are irrelevant. Its not about any of that, because moving those things between systems (eg my DSLR on the SCT vs Refractor) removes it as a variable. The rest is a system where you can choose two, but have to live with the third. You can get the same histogram (total light gathered) with any system, and stretch it from there.
Choose your image scale (focal length) and exposure time (F ratio, faster = lower) and live with the apeture you get.
Or
Choose your Aperture and image scale, and live with the F ratio (expsoure time) you get.
Etc.
If you want to image at 500mm with 16" scope, all you've got to do is find someone who can build you an F1.2 scope! It'll give you the same histogram as your 66mm F/7.5 apo (also 500mm) in just a fraction of the time.
-Brian
--------------------
Vixen 80EDSF
Stellarvue 102ED2
Orion Atlas 11 EQ-G
15" Astrosystems Telekit w/ Discovery Optics
Lust for something Larger
Lust for something Larger than that
Past Lovelies:
Oberwerk 20x100 binocs
Meade AR5
Meade LX10 8" SCT
All sold to a good home
|
Kolenka
Pooh-Bah
Reged: 06/01/08
Posts: 1009
Loc: Seattle Area, WA, USA
|
|
Quote:
I'm not sure what the points being made here are...
In order to get the same SNR as a 5 minute exposure at f/5 using an f/10 scope, you'd have to shoot for 20 minutes (4x the exposure). I know it won't be exactly the same because read noise is static and dark current noise is dependent only on exposure time and not aperture, but neither of these noise sources should be significant in a good camera at these exposure lengths.
Brightness per pixel != SNR
I'd actually expect the SNR in the 20 minute exposure to be better.
Say signal can be represented by an abstract number that represents the minutes we exposed. Signal we captured from the object at 20 minutes would be 20. Signal we captured from the object at 5 minutes would be 5. Aperture determines how many photons (signal) we capture. Focal length tells us how those photons get spread out.
So if we ignore read noise and dark current for a moment, we are left with shot noise, with the SNR being Signal over sqrt(Signal) to account for the shot noise.
As we expose longer, the SNR goes up, independent of focal length! So assuming we have the same aperture, the 20 minute exposure actually has an SNR of ~4.47 in our imaginary units, while the 5 minute exposure has an SNR of ~2.2 in our imaginary units. 4 times the exposure got us 2 times the SNR.
The caveat here is that read noise and dark noise are not totally insignificant. They are pixel noise that act totally independently of our system, and dark noise scales with exposure time. As skyglow overcomes the dark noise though, they both become less significant again (as the weakest part of our signal is now becoming a larger source of noise than the pixel noise).
But, as we move this back into the realm of numbers pulled off a sensor, we see two regions of performance from the sensor: One where pixel noise is dominant over shot noise and determines your SNR, and thus raw brightness of the pixel is important (which is where f-ratio has its biggest impact). The second region is where shot noise is dominant and determines your SNR.
The point where you leave the first region, and enter the second /is/ controlled by the f-ratio of the scope, since spreading the photons out pushes this point out to longer exposure times. As does large amounts of pixel noise. If a larger portion of the signal is pushed into each well, you overcome the pixel noise faster and move into the second region that much faster. Which both binning and faster f/ratios give you.
We also tend to spend a lot of time in the first region as beginners. The amount of signal we capture is pretty low, and amounts to sitting at a point not very high above the pixel noise (if at all). So we do see similar effects. But it doesn't mean we can extrapolate them out to longer exposures, as you start seeing different effects on SNR as you expose longer, since different forms of noise become dominant.
There is no hard line either, the 'inflection point' is simply the point where shot noise becomes larger than the pixel noise. You can go a bit further than that and still see effects of f-ratio easily, but they become weaker the longer you expose.
--------------------
Orion XX12 / Orion 80ED OTA / AT66ED
Nagler 7T6, 9T6, 13T6, 17T4, 26T5
Canon XS, TIS DMK 31AF03, AstroTrac TT320X
Northwest Astro Photoblog
|
Miguel Lopes
professor emeritus
Reged: 01/04/07
Posts: 694
Loc: Portugal
|
|
Quote:
Quote:
tiggere, you've left out the most important thing: the CCD!
That is why the term "fast" is not accurate for CCD's, especially those where you can bin the pixels!
I think the f ratio is important for image scale, optical defects/tolerances and backfocus. Everything else does not apply to the CCD world.
But that isn't entirely accurate either, and that was what spawned this whole discussion.
Say you have a graph of SNR versus exposure time. You would actually be able to split this graph into two parts, because SNR behaves differently in two regions of the graph. As you extend your time from 0 seconds, the noise is initially dominated by read noise, the slope of the line in this region is primarily determined by the f/ratio, or speed of the scope.
Once most of your signal is well above the read noise such that skyglow is now the dominant noise, you hit an inflection point of sorts, and the characteristics of the camera take over. The curve changes dramatically.
The only difference between an f/4 and f/8 scope of the same design and aperture, is that the f/4 will hit this inflection point sooner (closer to 0 seconds on the graph). Binning on an f/8 helps you reach this inflection point faster as well, but again, the camera takes over once you hit this inflection point.
Nocturnal, DSLRs are a pain to figure out here, partly because of their ISO and their noise. Because the dark current is so variable, DSLRs actually tend to push the inflection point out further compared to a cooled CCD, and will seem to support the idea that f-ratios matter when used. This is what I even encountered myself. Faster f-ratios gave me better DSLR shots, mostly because it was hard to get the signal well above the dark noise when shooting anything but open clusters.
Well, if we get into SNR...
Binning increases well depth, therefore increasing dynamic range (great side effect!) and in some situations reducing quantizing error.
So a f/8 binned will have much higher dynamic range than an f/4...
--------------------
Astrology is the science for ignorants. Astronomy is the science for those who feel ignorant. - Miguel Lopes
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
The important thing to remember is that while the number of photons that enter the aperture of any 8" telescope is the same regardless of focal ratio, it is only the photons that reach the camera's sensor that count in a photographic system. With an f/5 telescope, four times as many photons reach the sensor as with an f/10 telescope - the rest hit the OTA wall and are absorbed, reflected or diffracted, or pass right on through to the ground with a truss reflector. The same is true visually as well, but it's the exit pupil rather than the camera sensor that matters.
--------------------
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
|
Kolenka
Pooh-Bah
Reged: 06/01/08
Posts: 1009
Loc: Seattle Area, WA, USA
|
|
That doesn't quite seem correct for some reason.
Seems to me that the photons you lose are from the edges of the field. The subject in the field still has the same number of photons collected from it with the same aperture. As long as the subject fits in the FOV of the camera in both scopes, the aperture should collect the same number of photons from the subject, and the lost photons are from the edge of the field. Those photons you do collect are more spread out to fill the sensor.
Otherwise it would lead to a similar FOV, but dimmer result, rather than smaller FOV and dimmer result.
--------------------
Orion XX12 / Orion 80ED OTA / AT66ED
Nagler 7T6, 9T6, 13T6, 17T4, 26T5
Canon XS, TIS DMK 31AF03, AstroTrac TT320X
Northwest Astro Photoblog
|
Nocturnal
Pooh-Bah
Reged: 09/14/05
Posts: 1025
Loc: CT, USA
|
|
Quote:
That doesn't quite seem correct for some reason.
Seems to me that the photons you lose are from the edges of the field. The subject in the field still has the same number of photons collected from it with the same aperture. As long as the subject fits in the FOV of the camera in both scopes, the aperture should collect the same number of photons from the subject, and the lost photons are from the edge of the field. Those photons you do collect are more spread out to fill the sensor.
Otherwise it would lead to a similar FOV, but dimmer result, rather than smaller FOV and dimmer result.
Indeed. Hence my insistence that unless focal length (and thus image scale) is the same all attempts to compare F ratios are dubious at best. A scope A with the same focal length as scope B but twice the open aperture surface will collect twice as many photons at each cell. The image scale will be the same in both cases. The F ratio for scope A is smaller but it's the increased aperture that brings in the photons.
If you want detail, you need focal length. If you want high SNR, you need aperture. If you want both then you want large aperture and long focal length at the same time In the case of my C11 I gave up detail to gain signal strength by converting the F/10 to F/2. Photons that first didn't make it to the sensor now do. But they're packed closer together and thus I'm loosing detail.
--------------------
Best,
Sander
C11-HyperStar on Atlas EQ-Q driven by EQMOD
William Optics M110 With FR-III/TRF-2008
DS2090 guide scope
QHY-8, DSI-Pro and DSI cameras
watec 802h video camera with KIWI OSD
Astro stuff: http://www.tungstentech.com
My Astro Photos
|
gavinm
scholastic sledgehammer
Reged: 08/26/05
Posts: 804
Loc: Auckland New Zealand
|
|
Post deleted by gavinm
--------------------
Gavin
Mt Albert Grammar School Observatory
Auckland, New Zealand
http://www.mags.school.nz/astronomy/index.html
12" LX200R F6.8 AP
SBIG ST7-XME + CFW10
Moonlite SCT focuser w/ temp
Skywatcher Equinox ED80 Pro (ADM dovetail)
+ other stuff
|
lineman_16735
Tak-o-holic
Reged: 12/04/04
Posts: 2604
Loc: Central PA
|
|
Quote:
The important thing to remember is that while the number of photons that enter the aperture of any 8" telescope is the same regardless of focal ratio, it is only the photons that reach the camera's sensor that count in a photographic system. With an f/5 telescope, four times as many photons reach the sensor as with an f/10 telescope - the rest hit the OTA wall and are absorbed, reflected or diffracted, or pass right on through to the ground with a truss reflector. The same is true visually as well, but it's the exit pupil rather than the camera sensor that matters.
 I dunno about that? Seems to me as though you are confused about light cones. I don't mean that in a sarcastic way so please don't take it like that. This simply is not true though.
--------------------
Chris
A mount from Illinois
A scope from Japan
A camera from Cal-I-Fornia
A dog from Kentucky
A wife and kids from the "Twilight Zone"
The Geek Shed
|
lineman_16735
Tak-o-holic
Reged: 12/04/04
Posts: 2604
Loc: Central PA
|
|
Quote:
Quote:
That doesn't quite seem correct for some reason.
Seems to me that the photons you lose are from the edges of the field. The subject in the field still has the same number of photons collected from it with the same aperture. As long as the subject fits in the FOV of the camera in both scopes, the aperture should collect the same number of photons from the subject, and the lost photons are from the edge of the field. Those photons you do collect are more spread out to fill the sensor.
Otherwise it would lead to a similar FOV, but dimmer result, rather than smaller FOV and dimmer result.
Indeed. Hence my insistence that unless focal length (and thus image scale) is the same all attempts to compare F ratios are dubious at best. A scope A with the same focal length as scope B but twice the open aperture surface will collect twice as many photons at each cell. The image scale will be the same in both cases. The F ratio for scope A is smaller but it's the increased aperture that brings in the photons.
If you want detail, you need focal length. If you want high SNR, you need aperture. If you want both then you want large aperture and long focal length at the same time In the case of my C11 I gave up detail to gain signal strength by converting the F/10 to F/2. Photons that first didn't make it to the sensor now do. But they're packed closer together and thus I'm loosing detail.
Agreed for the most part 
--------------------
Chris
A mount from Illinois
A scope from Japan
A camera from Cal-I-Fornia
A dog from Kentucky
A wife and kids from the "Twilight Zone"
The Geek Shed
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
I guess I need to consider two different situations.
If you have a 10" scope with a focal length of 1000mm that would be an F-ratio of f/4. Let's consider the flux at one particular pixel of the camera, and say it is 20 electrons/second (I use electrons because it incorporates the quantum efficiency of the chip). If you cut the aperture in half, to 5" the f-ratio becomes f/8, and the flux at that pixel becomes only 5 electrons/second. You'd have to expose 4 times as long to record the same number of electrons and therefore acheive the same shot-noise SNR.
Now change the situation, instead of halving the aperture, double the focal length 2000mm. The f-ratio is f/8 again, but instead of less photons arriving at that pixel, what actually happens is the same number of photons arrive at an area equal to 4 pixels in a square. If the scope can resolve details that small, then it is possible that the illumination of each pixel will be slightly different. One could be only 2 electrons/sec, another 8/sec, another 4/sec, and the last 6/sec. In this situation the SNR of each pixel would be somewhat different. It could be that each pixel would get precisely 5 electrons/sec, in which case the situation is identical as above.
Quote:
Brightness per pixel != SNR
Not equal maybe, but the brightness value of a pixel is based entirely on how many photons it detects, and how much dark current builds up in it (and the bias charge I guess but it is insignificant). Since both of these processes are subject to shot noise, a pixel's brightness is directly correlated with SNR and SNR increases almost in direct proportion with brightness. The only reason it's an "almost" is read noise.
--------------------
-Arkalius
11" Celestron SCT on Orion Atlas EQ-G
Celestron 100ED Refractor
8" Zhumell Dobsonian Reflector
|
tiggere
super member
Reged: 02/22/08
Posts: 145
|
|
so basically this is why all the high end obs use RC's as there scope of choice...its got a very long focal length, plenty of aperture, and a moderate F ratio for more detail of the object they are shooting...added benefit being that they can fully illuminate a very large CCD...
I would rather have the resolution on the final product vs a faster SNR capture time as I can always add more shots...thats why I keep asking about the time differences between the two scopes...from what I have read your histogram should be around 1/4 to 1/3 over from the left...I might just test this next time I go out and see how long it takes to get there...I think I can manage 8 minute subs if I really try a good drift alignment with guiding (I'm a newbie to drift alignment) ...but if I'm understanding this correctly resolution can not be gotten back and the final image will suffer to a degree...
So a good compromise would be something in the middle of both like the new AT8RC...long focal length, plenty of aperture, and at F/8 your still not giving up alot of resolution...added bonus of a pretty flat field...granted its not a fix all scope as there are about 11 of the Messier objects that won't fit in the FOV...but who wants to own only one telescope  ...
--------------------
Celestron C6A-XLT
Atlas EQ_G with EQMOD / HNSKY
Logitech Dual Action Controller
Philips SPC900NC Webcam
Wilmington, NC
|
lineman_16735
Tak-o-holic
Reged: 12/04/04
Posts: 2604
Loc: Central PA
|
|
Quote:
So a good compromise would be something in the middle of both like the new AT8RC...long focal length, plenty of aperture, and at F/8 your still not giving up alot of resolution...
Resolution is a function of aperture. You can not increase resolution without increasing aperture.
FWIW I still stand by F ratio having nothing to do with SN. Aperture will only result in a SNR improvement.
--------------------
Chris
A mount from Illinois
A scope from Japan
A camera from Cal-I-Fornia
A dog from Kentucky
A wife and kids from the "Twilight Zone"
The Geek Shed
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
Quote:
Actually sky glow isn't noise, it's signal - but it contains noise (mostly shot noise in longer exposures) and since it is the least exposed part of the frame, the SNR of the sky glow due to shot noise is the poorest SNR in the image.
I may not have understand you correctly here Don, but sky glow is included in the entire frame. When we process our images, one of the first things we do is remove the sky glow signal by setting the black point to remove the gap to the left of the histogram. In other words, we basically subtract a single value from every pixel in the image. More below...
Earlier you said:
Quote:
Once you get past the read noise, the primary contributor to noise in DSLR and CCD photography is shot noise, which increases proportional to the square root of illumination (NOT exposure time). So the image of the extended object captured by the f/5 telescope has four times the signal and twice the shot noise, compared to the image captured by the f/10 telescope. Therefore the SNR due to shot noise is twice as high in the f/5 telescope compared to the f/10 telescope.
Yes but only if you ignore sky glow. Since the photons from sky glow obey the same rules of physics as the photons from the object we are interested in, the f/5 scope will also record four times the sky glow signal and twice the sky glow noise as the f/10 scope. However, since we basically throw away the sky glow signal as I mentioned above (four times as much with the f/5), we are still left with the sky glow noise, which is twice as high with the f/5 scope.
Therefore, in your example, after setting the black point on our image, the "effective" SNR per pixel of the f/5 will always be something less than twice that of the f/10 scope.
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
Quote:
Resolution is a function of aperture. You can not increase resolution without increasing aperture.
With imaging, resolution is a function of FL and pixel size up to a theoretical max that the scope can provide which is dictated by the aperature. So, if you are under rsampling with a particular scope & camera combo, you can increase your resolution without increasing aperature - within limits.
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
Quote:
Quote:
Quote:
tiggere, you've left out the most important thing: the CCD!
That is why the term "fast" is not accurate for CCD's, especially those where you can bin the pixels!
I think the f ratio is important for image scale, optical defects/tolerances and backfocus. Everything else does not apply to the CCD world.
But that isn't entirely accurate either, and that was what spawned this whole discussion.
Say you have a graph of SNR versus exposure time. You would actually be able to split this graph into two parts, because SNR behaves differently in two regions of the graph. As you extend your time from 0 seconds, the noise is initially dominated by read noise, the slope of the line in this region is primarily determined by the f/ratio, or speed of the scope.
Once most of your signal is well above the read noise such that skyglow is now the dominant noise, you hit an inflection point of sorts, and the characteristics of the camera take over. The curve changes dramatically.
The only difference between an f/4 and f/8 scope of the same design and aperture, is that the f/4 will hit this inflection point sooner (closer to 0 seconds on the graph). Binning on an f/8 helps you reach this inflection point faster as well, but again, the camera takes over once you hit this inflection point.
Nocturnal, DSLRs are a pain to figure out here, partly because of their ISO and their noise. Because the dark current is so variable, DSLRs actually tend to push the inflection point out further compared to a cooled CCD, and will seem to support the idea that f-ratios matter when used. This is what I even encountered myself. Faster f-ratios gave me better DSLR shots, mostly because it was hard to get the signal well above the dark noise when shooting anything but open clusters.
Well, if we get into SNR...
Binning increases well depth, therefore increasing dynamic range (great side effect!) and in some situations reducing quantizing error.
So a f/8 binned will have much higher dynamic range than an f/4...
This alludes to a point that I think causes some of the confusion on this subject.
There are a couple ways to look at SNR that the scope delivers (ignoring the camera for the moment):
1) Per unit area of sky covered (arc seconds square, arc minutes square what have you).
2) Per unit area at the image plane (microns square, pixels, what have you)
In the first case, aperture is the dominate factor (ignoring optical inefficiencies and such)
In the second case, FR is the dominant factor.
There are several important ramifications and relationships because of this. For example in the second case, if you increase your unit sample area by a factor of 4 times (say by binning 2x2) then you also increase your SNR.
This increase in SNR is such that if you took a shot with an 8" F/4 unbinned and an 8" F/8 binned 2x2 with the same camera & same length of exposure, your per effective pixel SNR will be the same as will be your image scale (resolution if you will) because each effective pixel has been hit by the same number of photons (and hence # 1 above). Of course the 8" F/8 will have 1/4 the FOV (by area of the sky) as the 8" F/4.
Another interesting aspect of this is that you can bin the image 2x2 in software after acquisition and achieve essentially the same result as binning with the camera (the only difference is binning in hardware has less read noise than binning in software).
You can also (in theory) get the same SNR with the 8" F/8 by using a camera with pixels twice as large (and all else being the same). Doing so results in the same image scale as the 8" F/4 with the camera with smaller pixels...
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
lineman_16735
Tak-o-holic
Reged: 12/04/04
Posts: 2604
Loc: Central PA
|
|
Quote:
Quote:
Resolution is a function of aperture. You can not increase resolution without increasing aperture.
With imaging, resolution is a function of FL and pixel size up to a theoretical max that the scope can provide which is dictated by the aperature. So, if you are under rsampling with a particular scope & camera combo, you can increase your resolution without increasing aperature - within limits.
Of course, a bad misplay on my part Dean. For some strange reason I was thinking of visual resolution and not imaging. I guess I lost my train of thought?
--------------------
Chris
A mount from Illinois
A scope from Japan
A camera from Cal-I-Fornia
A dog from Kentucky
A wife and kids from the "Twilight Zone"
The Geek Shed
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
Quote:
Quote:
The important thing to remember is that while the number of photons that enter the aperture of any 8" telescope is the same regardless of focal ratio, it is only the photons that reach the camera's sensor that count in a photographic system. With an f/5 telescope, four times as many photons reach the sensor as with an f/10 telescope - the rest hit the OTA wall and are absorbed, reflected or diffracted, or pass right on through to the ground with a truss reflector. The same is true visually as well, but it's the exit pupil rather than the camera sensor that matters.
You are basically saying that an F10 mirror has aberrations that reflect 3/4 of the photons out of the light cone, that an F5 mirror doesn't... That 3/4 photons don't even make the image plane in your average LX200... That if I add an F5 focal reducer right in front of the image plane, it somehow re-captures the photons that were previously being lost, absorbed, trussed-out..
This is the most ridiculous statement made in this whole focal ratio discussion. Find any reference to back up this statement, please.
I'm not talking about aberrations at all, Gavin. I'm talking about FOV. The faster scope with the same aperture has a wider FOV, in this case 4 times the area. So photons that reach the sensor on the faster scope don't reach the sensor on the slower scope, not due to aberrations but due to the FOV. In the case of f/10 vs. f/5, 3/4 of the photons entering the aperture from an evenly illuminated field don't reach the sensor (or the exit pupil in the visual case, assuming the same eyepiece is used in both scopes), though clearly they do enter the aperture. Where do you think they go? I thought you were a physics teacher.
--------------------
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 (05/06/09 01:50 PM)
|
gavinm
scholastic sledgehammer
Reged: 08/26/05
Posts: 804
Loc: Auckland New Zealand
|
|
Don. I owe you an apology
You just made me think of something that has been floating around upstairs, bumping into things and being annoying. My comment above was incorrect - I was looking at the situation from my (narrowminded?) physicists point of view. Some lateral thinking will be involved.
Again, sorry for my statement. I'll buy you a beer.
--------------------
Gavin
Mt Albert Grammar School Observatory
Auckland, New Zealand
http://www.mags.school.nz/astronomy/index.html
12" LX200R F6.8 AP
SBIG ST7-XME + CFW10
Moonlite SCT focuser w/ temp
Skywatcher Equinox ED80 Pro (ADM dovetail)
+ other stuff
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
No problem, Gavin. Apology accepted.
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
|
paulobao
sage
Reged: 06/26/07
Posts: 219
Loc: Porto, Portugal
|
|
Hi,
what's wrong at this simple thought: for a OTA with a diameter D, if you double your FL, your image diameter doubles at the focal plane and so, your surface is multiplied by 4, and the luminosity is divided by 4 (since the diameter D is the same so the number of photons entering the OTA's are the same too). ?
Regards,
paulo
--------------------
Clear skies
Paulo (MPC J11 and J15)
---------------------------------------------
Takahashi FS102NSV + Tak. reducer to f6.2
Takahashi FS60C + Tak. flattner to f6.2
Orion ATLAS EQ-G EQMOD
AP MaxBright diagonal
Aleph Lab 72º/40mm
UO HD 5, 6, 9 and 12 mm
UO 4 mm
KK Ortho 6 mm
QSI 532WS-M1
QHY6
QHY5 (CMOS)
Baader LRGBC CCD filters
Baader Ha-7nm, OIII-8nm, SII-8nm
Edited by paulobao (05/18/09 04:40 PM)
|
jgraham
Postmaster
Reged: 12/02/04
Posts: 6758
Loc: Dayton, Ohio
|
|
All I know is that it is a heck of a lot easier to image at f/5 than f/15 be it film, CCD, CMOS, or my built-in, face-mounted, biological photon detectors.
Have fun!
--------------------
-John
================================================
Homebuilt scopes from 4.25-16.5"
Meade LXD75-N6/SN6/SC8, DSX-90, ETX-60BB, ETX-125PE, DS-2130
Orion StarBlast, BinoViewers, Coronado PST
Rebel XT/XTi, DSI Pro (I, II, & III), DSI, LPI, Electronic Eyepiece, Phillips SPC900NC
Tasco 60mm Refractors
|
Kolenka
Pooh-Bah
Reged: 06/01/08
Posts: 1009
Loc: Seattle Area, WA, USA
|
|
Quote:
Hi,
what's wrong at this simple thought: for a OTA with a diameter D, if you double your FL, your image diameter doubles at the focal plane and so, your surface is multiplied by 4, and the luminosity is divided by 4 (since the diameter D is the same so the number of photons entering the OTA's are the same too). ?
The number of photons (overall) are the same as long as the aperture is the same, and the central obstruction is the same.
It gets tricky because this isn't the only thing that matters. Faster focal ratios do let you blow past the read noise faster, but if your exposures are already well past the read noise in both an f/5 and f/10 scope, your SNR is determined more by other attributes than your focal ratio. The focal ratio becomes insignificant.
The catch is that amateurs imaging galaxies with 2-5 minute subs really only get well past the read noise in the faster scopes, especially if you are running noisy cameras like DSLRs. It leads to the observations like jgraham hits:
Quote:
All I know is that it is a heck of a lot easier to image at f/5 than f/15 be it film, CCD, CMOS, or my built-in, face-mounted, biological photon detectors.
It isn't that he is wrong, but there is a realm (that we don't visit often with our subs) where this logic breaks down.
--------------------
Orion XX12 / Orion 80ED OTA / AT66ED
Nagler 7T6, 9T6, 13T6, 17T4, 26T5
Canon XS, TIS DMK 31AF03, AstroTrac TT320X
Northwest Astro Photoblog
|
russ_watters
Pooh-Bah
Reged: 11/24/04
Posts: 1277
Loc: Trappe, PA
|
|
Others alluded to this, but most people don't have a half dozen scopes and cameras from which to choose to match the subject they are shooting. Practical constraints get in the way.
So when someone talks about an object that fits in on the CCD regardless of the focal length....what if it doesn't? Doubling the focal length gives 4x the resolution, but if you can't fit the subject on the CCD anymore, you haven't helped yourself.
At the same time, I want good planetary imaging, which needs both focal length and aperture.
For exposure, I live in Pennsylvania, where clear skies are at a premium, so if I have to spend 4 days capturing an lrgb image, the odds of getting it successfully are not in my favor (I have some great luminance data of an M31 mosaic from a year ago and no color data...). Also, I use an EQ-G and 20 minute exposures aren't really a possibility at a long focal length anyway. Advantage: short focal ratio.
As someone else noted, telescopes don't have aperature dials on them (who would use one?), so you pick a combination of equipment that suits your needs and you deal with it. For my C11 and small chip DSI II Pro, a focal reducer is a must: Imaging at f/10 is too much focal length for my mount to be accurate with, too little fov for most objects, and not enough light for reasonable exposure lengths. A focal reducer helps deal with all those problems.
These discussions have some interesting theoretical value, but often they don't accurately represent the practical realities we're faced with.
--------------------
Equipment: Orion Atlas 11, ED80, DSI-C, DSI II Pro, Dell Inspiron Laptop.
www.russsscope.net
|
gavinm
scholastic sledgehammer
Reged: 08/26/05
Posts: 804
Loc: Auckland New Zealand
|
|
Quote:
The number of photons (overall) are the same as long as the aperture is the same, and the central obstruction is the same.
Needs clarification I think. Overall photons (from all sources) increase as focal ratio decreases, Overall photons from a point source remains constant, and depend on aperture only.
--------------------
Gavin
Mt Albert Grammar School Observatory
Auckland, New Zealand
http://www.mags.school.nz/astronomy/index.html
12" LX200R F6.8 AP
SBIG ST7-XME + CFW10
Moonlite SCT focuser w/ temp
Skywatcher Equinox ED80 Pro (ADM dovetail)
+ other stuff
|
Kolenka
Pooh-Bah
Reged: 06/01/08
Posts: 1009
Loc: Seattle Area, WA, USA
|
|
Quote:
Quote:
The number of photons (overall) are the same as long as the aperture is the same, and the central obstruction is the same.
Needs clarification I think. Overall photons (from all sources) increase as focal ratio decreases, Overall photons from a point source remains constant, and depend on aperture only.
For the intent of getting signal from an object (extended or otherwise), aperture is the main means for collecting photons. Focal length determines how they are spread out. You only lose signal due to focal length changes if the focal length change means the object no longer fits in the FOV of the CCD sensor.
--------------------
Orion XX12 / Orion 80ED OTA / AT66ED
Nagler 7T6, 9T6, 13T6, 17T4, 26T5
Canon XS, TIS DMK 31AF03, AstroTrac TT320X
Northwest Astro Photoblog
|
gavinm
scholastic sledgehammer
Reged: 08/26/05
Posts: 804
Loc: Auckland New Zealand
|
|
Agreed - that was my second point. Photons from an individual source only depends on clear aperture.
--------------------
Gavin
Mt Albert Grammar School Observatory
Auckland, New Zealand
http://www.mags.school.nz/astronomy/index.html
12" LX200R F6.8 AP
SBIG ST7-XME + CFW10
Moonlite SCT focuser w/ temp
Skywatcher Equinox ED80 Pro (ADM dovetail)
+ other stuff
|
paulobao
sage
Reged: 06/26/07
Posts: 219
Loc: Porto, Portugal
|
|
Photons are translated in energy. The energy you collect are only clear aperture dependent. And when you double the FL (same as saying your f/d doubles) you spread the same amount of energy by a 4x larger surface. This is something nobody can deny!
So, someone showed here 2 pictures from the same instrument: one at f/3.9 and the other at 12.something. They are both ok and equal (yeah right). The problem here as everyone knows is that I bet the one at f/12. something is well stretched, at the limit and the one at f3.9 is well understretched. You could obtain much more information of the one at f/3.9 if you continues to stretch it!
I bet I can put here 2 photos taken with the same instruments (say, my FS102 and my QSI532WS-M1 with a HII filter) one of 10 min and the other of 20 min and both seems exactly equal. So there is no point of increasing exposure time, right? Wrong as you know! So why they will appear the same? Because the stretching of the histos are different!
Ok, I will make this simple experiment when I have time: since my chip is a NABG one (KAF3200ME), I will take a series of exposure from a light source (star) at f/6 and f/8. Then I will determine at what time the star starts blooming at f/6 and at f/8. If they will start blooming at the same exposure time then the f/d should be not important, otherwise we will need to consider the f/d.
--------------------
Clear skies
Paulo (MPC J11 and J15)
---------------------------------------------
Takahashi FS102NSV + Tak. reducer to f6.2
Takahashi FS60C + Tak. flattner to f6.2
Orion ATLAS EQ-G EQMOD
AP MaxBright diagonal
Aleph Lab 72º/40mm
UO HD 5, 6, 9 and 12 mm
UO 4 mm
KK Ortho 6 mm
QSI 532WS-M1
QHY6
QHY5 (CMOS)
Baader LRGBC CCD filters
Baader Ha-7nm, OIII-8nm, SII-8nm
Edited by paulobao (05/19/09 04:58 AM)
|
Kolenka
Pooh-Bah
Reged: 06/01/08
Posts: 1009
Loc: Seattle Area, WA, USA
|
|
Quote:
So, someone showed here 2 pictures from the same instrument: one at f/3.9 and the other at 12.something. They are both ok and equal (yeah right). The problem here as everyone knows is that I bet the one at f/12. something is well stretched, at the limit and the one at f3.9 is well understretched. You could obtain much more information of the one at f/3.9 if you continues to stretch it!
If that information is there to begin with, and can be extracted without totally distorting the rest of the image. Or that it is above the read noise itself.
Quote:
I bet I can put here 2 photos taken with the same instruments (say, my FS102 and my QSI532WS-M1 with a HII filter) one of 10 min and the other of 20 min and both seems exactly equal. So there is no point of increasing exposure time, right? Wrong as you know! So why they will appear the same? Because the stretching of the histos are different!
This seems to miss the point of the argument of the f-ratio myth. I've not seen anyone say "There is no point of increasing exposure time past X". The reality is that up to a certain exposure time for a scope, SNR is determined by the focal ratio. Once your exposure times put the signal well above the read noise (and this exposure length is determined by f-ratio), SNR is more determined by the camera characteristics. Shot noise will still decrease with time, and so on...
This is not brightness, but SNR which is a better objective measure of collected data because we /do/ have to stretch.
Quote:
Ok, I will make this simple experiment when I have time: since my chip is a NABG one (KAF3200ME), I will take a series of exposure from a light source (star) at f/6 and f/8. Then I will determine at what time the star starts blooming at f/6 and at f/8. If they will start blooming at the same exposure time then the f/d should be not important, otherwise we will need to consider the f/d.
This again is a measure of pure brightness, and doesn't seem to effectively measure the real SNR effects you will see on extended objects. A faster focal ratio will be 'brighter' at the same exposure length, but that doesn't guarantee less noise (because of factors like shot noise/etc).
--------------------
Orion XX12 / Orion 80ED OTA / AT66ED
Nagler 7T6, 9T6, 13T6, 17T4, 26T5
Canon XS, TIS DMK 31AF03, AstroTrac TT320X
Northwest Astro Photoblog
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
Paulo, Since you're willing to experiment, try shooting the same star with both setup's, but use a short enough exposure that neither scope saturates the sensor, take 50-100 of each, and the stack those. Then measure the star's value over skyfog, and measure SNR. AIP4Win has some SNR tools, I'm sure some of the other processing tools do as well. In the Northern Hemisphere, Arcturus has a 9th and 10th Mag star close by, which in my system doesn't saturate the sensor in a 1 sec sub. Find such a target, shoot it with both scopes and we'll know.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
A star is a bad choice for an experiment for this purpose. Stars are point sources and as such the photons you get for them is based on aperture only, and not affected by focal length. The only thing focal length will affect is the apparant size of the airy disc on the sensor. This can be meaningful if the resloution of the system is high enough...
--------------------
-Arkalius
11" Celestron SCT on Orion Atlas EQ-G
Celestron 100ED Refractor
8" Zhumell Dobsonian Reflector
|
paulobao
sage
Reged: 06/26/07
Posts: 219
Loc: Porto, Portugal
|
|
A galaxy core will be used to measure the SNR.
I will make the experiment as soon I have good weather here at my obervatory place in Portugal. Believe me, I would prefer to work without the reducer all the time.
--------------------
Clear skies
Paulo (MPC J11 and J15)
---------------------------------------------
Takahashi FS102NSV + Tak. reducer to f6.2
Takahashi FS60C + Tak. flattner to f6.2
Orion ATLAS EQ-G EQMOD
AP MaxBright diagonal
Aleph Lab 72º/40mm
UO HD 5, 6, 9 and 12 mm
UO 4 mm
KK Ortho 6 mm
QSI 532WS-M1
QHY6
QHY5 (CMOS)
Baader LRGBC CCD filters
Baader Ha-7nm, OIII-8nm, SII-8nm
Edited by paulobao (05/20/09 06:04 AM)
|
Galaxyhunter
Pooh-Bah
Reged: 01/02/06
Posts: 1251
|
|
Quote:
A star is a bad choice for an experiment for this purpose. Stars are point sources and as such the photons you get for them is based on aperture only, and not affected by focal length.
Arkalius, Isn't a star only considered a "Point Source" if it is only registered on a single pixel
--------------------
Carl
My lousy skies at Hawkeye Observatory
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
Quote:
A star is a bad choice for an experiment for this purpose. Stars are point sources and as such the photons you get for them is based on aperture only, and not affected by focal length. The only thing focal length will affect is the apparant size of the airy disc on the sensor. This can be meaningful if the resloution of the system is high enough...
You might be right, but I'm not sure. I think it shouldn't be any different shooting a 6th Mag star or an extented object with a surface brightness of 6th Mag. I know that stars from my C8 are bigger than stars from my Onyx.
But, if you did shoot a galaxy core, there should also be some stars to test as well to see if that makes a difference or not.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
Quote:
Arkalius, Isn't a star only considered a "Point Source" if it is only registered on a single pixel
No... It's a point source because stars are too small to resolve any apparant size with a telescope. They do produce a diffraction pattern that itself has an apparant size. It is the Airy disk (and seeing effects, as well as halation) that makes stars occupy multiple pixels on the sensor.
I actually graphed the relationship between focal ratio and SNR, keeping all other variables constant. The graph has the shape of a function y = 1 / x^a where a is generally larger than 1 but much less than 2. Focal ratio is x and y is the SNR. So, a halving of focal ratio (all other things being equal) would produce something more than a doubling but probably a lot less than a quadrupling of the SNR.
--------------------
-Arkalius
11" Celestron SCT on Orion Atlas EQ-G
Celestron 100ED Refractor
8" Zhumell Dobsonian Reflector
|
gavinm
scholastic sledgehammer
Reged: 08/26/05
Posts: 804
Loc: Auckland New Zealand
|
|
Quote:
Arkalius, Isn't a star only considered a "Point Source" if it is only registered on a single pixel
Point source means unresolvable. A star is a point source in any situation (for our purposes)
--------------------
Gavin
Mt Albert Grammar School Observatory
Auckland, New Zealand
http://www.mags.school.nz/astronomy/index.html
12" LX200R F6.8 AP
SBIG ST7-XME + CFW10
Moonlite SCT focuser w/ temp
Skywatcher Equinox ED80 Pro (ADM dovetail)
+ other stuff
|
Jared
Carpal Tunnel
Reged: 10/11/05
Posts: 2531
Loc: Piedmont, California, U.S.
|
|
It seems like many of us are not understanding the difference between per pixel SNR and object SNR. The object SNR is not affected by changes in focal ratio--either through use of a barlow/focal reducer or through choosing a camera with larger or smaller pixels or through binning. In all cases, the object SNR is the same (ignoring read noise). Object SNR is determined purely by aperture.
However, per pixel SNR is affected by focal length (and therefore, since we are holding aperture constant, focal ratio). If you cut your focal ratio in half, you will be improving your per pixel SNR by a factor of two (four times the signal at each site divided by twice the noise yields twice the per pixel SNR).
As Dean correctly pointed out, you could accomplish this by inserting a focal reducer, binning, or after the fact--in software--by applying a two pixel Gaussian blur filter. These are all functionally equivalent.
Effectively, you can trade resolution for improved smoothness in your image by adjusting focal length or pixel size (using focal reducers and/or binning) or by running noise reduction software--essentially a selective blurring filter. This assumes you are undersampled, of course--that your chip isn't already recording all the resolution the sky conditions will allow. It also assumes that read noise is negligible.
So why doesn't this fit our real world experiences? Because for many of us read noise isn't negligible. With relatively short subexposures, read noise is still a significant contributor to overall noise. Faster scopes do allow shot noise to quickly swamp read noise.
--------------------
- Jared Willson
- TMB 152 f/8 Apochromat
- Fluorostar FLT-110 w/ TEC optics
- Stellarvue SV80S
- Astro-Physics Mach1 GTO
- Takahashi Teegul SP Mount
- STL-11000
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
Quote:
It seems like many of us are not understanding the difference between per pixel SNR and object SNR.
I would say Object SNR would be an average of the SNR of all the pixels that make up the object in the image... does that work? I'm not sure there's any other useful definition.
Quote:
The object SNR is not affected by changes in focal ratio--either through use of a barlow/focal reducer or through choosing a camera with larger or smaller pixels or through binning. In all cases, the object SNR is the same (ignoring read noise). Object SNR is determined purely by aperture.
If per-pixel SNR is changed by altering the focal length (as you mention in the quote below), and we define object SNR as an average of the SNR of the pixels that make it up, then it follows that object SNR would be reduced along with the per-pixel SNR.
Quote:
However, per pixel SNR is affected by focal length (and therefore, since we are holding aperture constant, focal ratio). If you cut your focal ratio in half, you will be improving your per pixel SNR by a factor of two (four times the signal at each site divided by twice the noise yields twice the per pixel SNR).
The math isn't that simple actually. Since a change in focal ratio doesn't affect read noise or dark current noise, those items remain constant in the equation. Because of these elements, a halving of focal ratio will actually do something more than double the SNR.
The amount of signal you get depends on focal ratio, (aperture only for point sources), pixel pitch (linear size), and exposure time. Halving pixel pitch has the same effect as doubling focal ratio, signal-wise. This is why oversampling can be considered such a heinous crime. You are sacrificing sensitivity for nothing. Undersampling is at least a trade-off situation-- more signal at the cost of resolution.
--------------------
-Arkalius
11" Celestron SCT on Orion Atlas EQ-G
Celestron 100ED Refractor
8" Zhumell Dobsonian Reflector
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
Quote:
Quote:
It seems like many of us are not understanding the difference between per pixel SNR and object SNR.
I would say Object SNR would be an average of the SNR of all the pixels that make up the object in the image... does that work? I'm not sure there's any other useful definition.
What works for me is to think of SNR in units of the sky (arc secs or minutes square) and in area at the image plane (mm or um square or pixels). It also helps me to think of SNR as delivered by the scope at the image plane and SNR as the camera records it.
There are several reasons for this. The scope doesn't know or care what camera is attached to it or the size of its pixels, QE, read noise etc. Take the same scope, same object and exposure time but change cameras and the resulting per pixel SNR changes for reasons that have nothing to do with the scope.
Quote:
Quote:
The object SNR is not affected by changes in focal ratio--either through use of a barlow/focal reducer or through choosing a camera with larger or smaller pixels or through binning. In all cases, the object SNR is the same (ignoring read noise). Object SNR is determined purely by aperture.
If per-pixel SNR is changed by altering the focal length (as you mention in the quote below), and we define object SNR as an average of the SNR of the pixels that make it up, then it follows that object SNR would be reduced along with the per-pixel SNR.
If you ignore the camera and consider the SNR per unit of the sky (arc secs or arc mins square) the scope delivers at the image plane, then the SNR is unchanged by adding a reducer/barlow (assuming no light loss). However, the SNR is changed per unit area of the image plane (square um, mm, inches, feet whatever) - but that's a constraint imposed by the camera.
Quote:
Quote:
However, per pixel SNR is affected by focal length (and therefore, since we are holding aperture constant, focal ratio). If you cut your focal ratio in half, you will be improving your per pixel SNR by a factor of two (four times the signal at each site divided by twice the noise yields twice the per pixel SNR).
The math isn't that simple actually. Since a change in focal ratio doesn't affect read noise or dark current noise, those items remain constant in the equation. Because of these elements, a halving of focal ratio will actually do something more than double the SNR.
But the scope doesn't have read noise or dark current - the camera does!
Also (and off the point I'm trying to make) there's sky glow. While reducing the FR increases the number of photons per unit area per unit time from the object, it also increases the number of photons from the sky glow as well. We tend to throw away the signal from the sky glow by setting the black point when we process the image, but we can't throw away the noise from the sky glow so this effectively reduces the increase in SNR compared to if we ignore (or had no) sky glow.
Quote:
The amount of signal you get depends on focal ratio, (aperture only for point sources), pixel pitch (linear size), and exposure time. Halving pixel pitch has the same effect as doubling focal ratio, signal-wise. This is why oversampling can be considered such a heinous crime. You are sacrificing sensitivity for nothing. Undersampling is at least a trade-off situation-- more signal at the cost of resolution.
The scope doesn't have pixel pitch - that's a constraint imposed by the camera as is over/under sampling.
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
Without some kind of detector, be it your eyeball or a camera sensor, you don't have an SNR. You cannot disconnect the camera from the discussion of SNR.
It also doesn't need to be all that complicated. It's simply the ratio of an expected signal magnitude over the RMS error magnitude. It's a statistical measurement, and statistical measurements deal with samples. When it comes to a camera sensor, the sample is the pixel. The SNR of the pixels ultimately is based on how many photons it detects. How many it detects depends on the pixel's QE and how many photons struck it. The number of photons that strike it is dependent on focal ratio of the optical system, the size of the pixel, the amount of time spent detecting photons, and how many photons are coming from the scene at that point in the image. All of these quantities factor into measuring the SNR. Considering the SNR independently of some form of detector is meaningless.
--------------------
-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
|
|
Here's my take on this.
Aperture controls how much light enters the optics.
Focal length of the optics control the image scale, size, and area of the focal plane. Your camera controls whether you're critically sampled, oversampled or undersampled. When you're undersampled you are imaging a point source, as that's what a point source really is. However if you're oversampled your object is just spread over more than one pixel and it's really an extended object, even if it's a star.
The ratio of the image scale of the light coming in to the scale of the focal plane controls how much the optics intensifies the incoming light
I've been trying to measure the sensitivity of my system, and even with a 1 sec exposure Arcturus is ~ 800 pixels. And while I can't accurately measure arcturus, I can some fainter stars, and the peak pixel value doesn't work to derive the magnitude, you have to add up the signal for all of the pixels, that seems to work.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
Edited by Qkslvr (05/22/09 07:56 AM)
|
Jared
Carpal Tunnel
Reged: 10/11/05
Posts: 2531
Loc: Piedmont, California, U.S.
|
|
Quote:
Quote:
It seems like many of us are not understanding the difference between per pixel SNR and object SNR.
I would say Object SNR would be an average of the SNR of all the pixels that make up the object in the image... does that work? I'm not sure there's any other useful definition.
Actually, I don't think that definition works for me because it takes the limitations imposed by the camera and builds them back into a value that is intended to be independent of the detector. Object SNR is supposed to be a measure of the light coming into the scope, independent of equipment. I know you argue that it is meaningless. Instead, I would suggest that it is meaningful but insufficient by itself to determine image quality.
Quote:
Quote:
The object SNR is not affected by changes in focal ratio--either through use of a barlow/focal reducer or through choosing a camera with larger or smaller pixels or through binning. In all cases, the object SNR is the same (ignoring read noise). Object SNR is determined purely by aperture.
If per-pixel SNR is changed by altering the focal length (as you mention in the quote below), and we define object SNR as an average of the SNR of the pixels that make it up, then it follows that object SNR would be reduced along with the per-pixel SNR.
I don't agree with your definition of object signal to noise--exactly because it would have the result you describe. Object signal to noise is completely independent of camera, and therefore of focal ratio. It depends only on aperture (and central obstruction, of course). Certainly, you can make an argument that object SNR does not, by itself, determine image quality. That is true.
Quote:
Quote:
However, per pixel SNR is affected by focal length (and therefore, since we are holding aperture constant, focal ratio). If you cut your focal ratio in half, you will be improving your per pixel SNR by a factor of two (four times the signal at each site divided by twice the noise yields twice the per pixel SNR).
The math isn't that simple actually. Since a change in focal ratio doesn't affect read noise or dark current noise, those items remain constant in the equation. Because of these elements, a halving of focal ratio will actually do something more than double the SNR.
True. I was ignoring read and thermal noise intentionally for simplicity.Quote:
The amount of signal you get depends on focal ratio, (aperture only for point sources), pixel pitch (linear size), and exposure time. Halving pixel pitch has the same effect as doubling focal ratio, signal-wise. This is why oversampling can be considered such a heinous crime. You are sacrificing sensitivity for nothing. Undersampling is at least a trade-off situation-- more signal at the cost of resolution.
Here we agree. When undersampling, you are trading spatial resolution for per pixel SNR. Oversampling does not accomplish anything--though we could argue further about where oversampling starts... Of course, you can get rid of your oversampling after image capture, but you still would have the negative aspect of the increased read noise.
--------------------
- Jared Willson
- TMB 152 f/8 Apochromat
- Fluorostar FLT-110 w/ TEC optics
- Stellarvue SV80S
- Astro-Physics Mach1 GTO
- Takahashi Teegul SP Mount
- STL-11000
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
Quote:
I don't agree with your definition of object signal to noise--exactly because it would have the result you describe. Object signal to noise is completely independent of camera, and therefore of focal ratio.
Jared,
It can't be just aperture, you have to also account for focal length, f/ratio is what accounts for an increase (or decrease) of intensity where you increase aperture and maintain equal image scales (focal length). Otherwise it's just a piece of window glass.
IMO SNR is only meaningful for a detected signal. You can model any image as a specific point source for each pixel, and it's value will be a sum of all of the signals present, "Signal", skyfog, dark noise and read noise.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
Quote:
Without some kind of detector, be it your eyeball or a camera sensor, you don't have an SNR. You cannot disconnect the camera from the discussion of SNR.
Not really.
Quote:
It's a statistical measurement, and statistical measurements deal with samples. When it comes to a camera sensor, the sample is the pixel. The SNR of the pixels ultimately is based on how many photons it detects. How many it detects depends on the pixel's QE and how many photons struck it. The number of photons that strike it is dependent on focal ratio of the optical system, the size of the pixel, the amount of time spent detecting photons, and how many photons are coming from the scene at that point in the image. All of these quantities factor into measuring the SNR. Considering the SNR independently of some form of detector is meaningless.
Ah - but the number of photons that strike the sensor has a Poisson distribution which means you have an uncertainty (noise) of the square root of the photon count.
So if you have 100 photons that strike the sensor, you have an uncertainty of 10 ( the sqrt(100) ) and thus a SNR of 10 ( 100/10 ). That's true whether you are talking about light striking a CCD sensor or a piece of plywood.
In other words, you have a source of noise that has nothing whatsoever to do with the detector - that's far from being meaningless.
Quote:
It also doesn't need to be all that complicated.
Personally, I don't see it as all that complicated. For me, thinking of SNR at the image plane of the scope in both units of the sky as well as units of area as well as SNR as produced by the camera has helped me to underdressed why discussion of this subject always goes around in circles.
I'm not saying that FR does not influence pixel SNR for a given pixel size, I agree with that. What I'm saying is that pixel size is a variable in the equation.
Think of it this way - you take a 4" FSQ at F/5 and whatever camera you care to use and I'll take a 20" RCOS at F/9 and whatever camera I care to use and we both image M51 for the same length of time. I'll bet that with the 20" at F/9 I can achieve both higher SNR and higher resolution than you can with the 4" at F/5.
The reason I can say that is because not matter what camera you choose with the 4" FSQ, you are going to be limited by the SNR per unit of the sky. The 20" RCOS will also be limited by the SNR per unit sky, but it will be a much higher limit. That means that with the 20" RCOS I have a much more room to trade resolution for SNR than you would with the 4" FSQ. And there's 1 reason for that - aperture.
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
The RCOS has 25x the aperture, and ~9x the focal length, which would give a little over 1 Mag increased intensity. But the fov is also about 1/9 as big. Okay for M51, not so good for M31. Or you have to buy a really big/expensive detector.
That also shows that FR isn't a complete answer.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
Quote:
The RCOS has 25x the aperture, and ~9x the focal length, which would give a little over 1 Mag increased intensity. But the fov is also about 1/9 as big. Okay for M51, not so good for M31. Or you have to buy a really big/expensive detector.
Well, I did say any camera I choose
http://www.cfa.harvard.edu/~bmcleod/MegacamCCDInstallation/IMGP0613.htm
But seriously, there certainly are many factors and considerations (including practicality), otherwise everyone would be using the same scope and camera.
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
Jared
Carpal Tunnel
Reged: 10/11/05
Posts: 2531
Loc: Piedmont, California, U.S.
|
|
Quote:
Quote:
I don't agree with your definition of object signal to noise--exactly because it would have the result you describe. Object signal to noise is completely independent of camera, and therefore of focal ratio.
Jared,
It can't be just aperture, you have to also account for focal length, f/ratio is what accounts for an increase (or decrease) of intensity where you increase aperture and maintain equal image scales (focal length). Otherwise it's just a piece of window glass.
IMO SNR is only meaningful for a detected signal. You can model any image as a specific point source for each pixel, and it's value will be a sum of all of the signals present, "Signal", skyfog, dark noise and read noise.
Looks like we will just have to disagree on this one. I believe that object SNR is meaningful independent of the detector because it tells me how much total exposure time I am going to need based on my aperture, sky conditions, and subject in order to get the image quality I am looking for.
Does it give me everything I need? Of course not. Focal ratio, pixel pitch, read noise, and dark current all play a role as well. But that doesn't mean object SNR is meaningless.
There is, in fact, a SNR inherent in the photons streaming into the telescope that is independent of the detector. The light from my subject follows a Poisson distribution (as Dean mentioned), and the sky glow follows the same distribution. This distribution is independent of focal ratio and focal length. In fact, it is independent of QE, pixel size, read noise, and dark current as well.
--------------------
- Jared Willson
- TMB 152 f/8 Apochromat
- Fluorostar FLT-110 w/ TEC optics
- Stellarvue SV80S
- Astro-Physics Mach1 GTO
- Takahashi Teegul SP Mount
- STL-11000
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
Quote:
I believe that object SNR is meaningful independent of the detector because it tells me how much total exposure time I am going to need based on my aperture, sky conditions, and subject in order to get the image quality I am looking for.
It really sounds like you are talking about object flux. SNR is essentially the inverse to the amount of error in the measurement of something. SNR is a quality of the actual measurement of something, not the thing itself. An object's brightness will definitely give you the last part of the puzzle in regards to how much exposure you need, but that doesn't mean the object has some kind of SNR.
Quote:
There is, in fact, a SNR inherent in the photons streaming into the telescope that is independent of the detector. The light from my subject follows a Poisson distribution (as Dean mentioned), and the sky glow follows the same distribution.
There is an inherent source of noise in the stream of photons due to the stochastic nature of light, which, when measured, will lead to the potential for calculating an SNR, but the light itself doesn't have a built-in SNR.
Quote:
This distribution is independent of focal ratio and focal length. In fact, it is independent of QE, pixel size, read noise, and dark current as well.
All of this is true, but you forgot to mention that the Poisson distrbution requires a finite time. And, in order to know what the average rate of photons is to calculate the distribution, you'd need to know things like focal ratio and pixel size. If you are using a camera to measure these photons, then things like read noise and dark current factor into the SNR at that point. If you want to get really abstract, we can avoid discussion of a physical detector. But, you cannot avoid the fact that the SNR is based on the measurement of a quantity.
To measure SNR, you have to measure a finite amount of something. In this case, we're talking about photons. The most effective way of doing that is multiplying a photon flux value by a time and area. Flux is a measurement of the number of things that pass through or occur in a particular finite area per unit of time.
Object brightness and aperture will give you a start on measuring the flux from the object. Focal length, and the area at the image plane you are discussing will give you the rest. To make a finite measurement of photons, you need to pick a specific amount of time. So, the things we need to make an SNR measurement are focal ratio (which contains aperture and focal length), a finite time, and a finite area at the image plane. If you wanted to choose an area equal to the area of the object itself, and multiply that by an average object flux, you'd end up with a rather high rate of photons typically, and could reach a large SNR in a rather short period of time. But, the only data you'd have is how bright the object is, and would know nothing about the details of it's appearance.
To figure that part out, you have to measure brightness in various spots on the target and generate an image out of it. These spots will be smaller than the target itself of course, and the target flux in each discrete area will vary based on the appearance and shape of the target. The smaller the measured area becomes, the lower the photon arrival rate will be, and the lower the SNR will be when the measurement time remains fixed.
As you can see, to get a measurement of SNR, we need to consider a finite time of measurement, and the finite area of the image plane that we are measuring. These are values that cannot be extracted out of a discussion that focusses only on the telescope and the object. You need to consider the parameters of some kind of measurement. A camera allows you to easily specify an area, and the act of taking a photo implies a finite time of measurement.
--------------------
-Arkalius
11" Celestron SCT on Orion Atlas EQ-G
Celestron 100ED Refractor
8" Zhumell Dobsonian Reflector
Edited by Arkalius (05/22/09 07:43 PM)
|
Jared
Carpal Tunnel
Reged: 10/11/05
Posts: 2531
Loc: Piedmont, California, U.S.
|
|
After reading through your post carefully, I think maybe we are saying the same thing but with slightly different semantics and different points of emphasis.
Correct me if I'm wrong, but I think we both agree with the following:- There is a source of error in the photons flowing into the telescope that is independent of focal ratio and detector; the effects of this error are dependent on aperture alone (since the error goes up as the square root of the total count of photons from the subject, and the total number of photons captured increases with aperture alone)
- SNR in the final image is dependent on many factors including aperture, read noise, thermal noise, pixel size, binning, sky glow, and total exposure time
- As long as you are not oversampling, it is possible to trade resolution for better SNR by either choosing a faster focal ratio or a camera with larger pixels--either solution should work equally well ignoring field of view considerations and edge aberrations, etc.
- Focal ratio is a significant factor in the SNR in the image itself since it affects per pixel SNR
I think I phrased all that in a way that we can both agree, yes? I'd hate to let this thread end without the beginners who have managed to wade through it left with some points on which there is no fundamental disagreement.
--------------------
- Jared Willson
- TMB 152 f/8 Apochromat
- Fluorostar FLT-110 w/ TEC optics
- Stellarvue SV80S
- Astro-Physics Mach1 GTO
- Takahashi Teegul SP Mount
- STL-11000
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
That pretty much sums it up, though you have to remember that even though it is true that the total number of photons from the target is dependent only on clear aperture, measuring SNR requires measuring the photons. Since a detector is going to be of a specific size, how those photons from the target are spread out becomes important, so focal length has to factor in. Even stars are affected by focal length because FL affects the size of the Airy disk.
I think the easy bottom line is that in general, a faster focal ratio will improve SNR all other things being equal. You can also improve SNR by using a camera with larger pixels (or binning) but at the cost of resolution.
--------------------
-Arkalius
11" Celestron SCT on Orion Atlas EQ-G
Celestron 100ED Refractor
8" Zhumell Dobsonian Reflector
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Quote:
Quote:
... Maybe it's different with CCD's, but until I hear it from an optical physicist or someone like that I'm skeptical...
You may well be hearing it from physicists, but some other people still think they know better..
I don't know any credentialed physicist or optical scientist who doesn't consider f/ratio just as fundamental now, with ccd, as it ever was, with film.
For extended objects, the adu in each pixel goes as the inverse square of the f/ratio, independent of aperture. As a result, assuming shot noise, the SNR goes as the inverse of the f/ratio - independent of aperture.
With smaller aperture, same detector, and same f/ratio, you lose resolution, but the per-pixel SNR is the same.
For stars, the SNR expressions are much more complicated and are largely determined by aperture - but not entirely.
Schroeder's book on astronomical optics describes the equations. I don't know any text or journal article that describes the f/ratio as a myth or that somehow f/ratio is no longer relevant with ccd's.
Frank
|
lernerda
professor emeritus
Reged: 01/21/07
Posts: 544
Loc: New Jersey 40.0 N 74.9 W
|
|
OK, let's look at something a little different.
Same CCD, 4":F10 vs 8":F5. Both have a focal length of 40".
Using the same ccd (assume a small CCD less than 1" diagonal), the image "size" should be exactly the same. M51 would look the same in both images.(I know the resolution will be different as the airy disk is defined by the diameter of the scope)
Here is the question. What about the "brightness" of the image? Will the exposure times be the same?
(Looking at it another way, IF you could line up the image exactly the the same, so the same pixels from each scope would seeing the same part of the object, would the number of photons hitting the repective pixels for a given amount of time be the same for each scope)
--------------------
David
www.seeingafterdark.com
Tak EM200, SV ED102, WO Z66ED, 0.8 Reducer/Flattener
Meade DSI-3 Mono w/ Orion's Skyglow Imaging & LRGB filters, DSI-2 for guiding, Philips SPC900NC
25x100 Burgess, 20X80 Konus Binocs, & homemade P-mount
Edited by lernerda (06/05/09 10:18 AM)
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
With the same length exposure, the 8" image will be both brighter and have a higher SNR.
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Quote:
Same CCD, 4":F10 vs 8":F5. Both have a focal length of 40".
Using the same ccd (assume a small CCD less than 1" diagonal), the image "size" should be exactly the same. M51 would look the same in both images.(I know the resolution will be different as the airy disk is defined by the diameter of the scope)
Here is the question. What about the "brightness" of the image? Will the exposure times be the same?
For the galaxy itself - not the stars in the scene - you can ignore aperture and just go by f/ratio in terms of brightness or adu in each pixel. The f/5 will be four (2^2) times brighter than the f/10.
This would also be true if you used a 200" aperture f/10 - the 8" f/5 would have 4 times the adu for the extended glow from the galaxy.
Both film and ccd would be "brighter" with the 8" - but the galaxy would be much larger with the 200". No myths apply - this stuff is as old as the hills.
Frank
|
HunterofPhotons
sage
Reged: 04/26/08
Posts: 266
Loc: Rhode Island, USA
|
|
Quote:
....... I don't know any text or journal article that describes the f/ratio as a myth or that somehow f/ratio is no longer relevant with ccd's.
Frank
I think that you've missed the point. The only thing that Stan is saying is that suppose that you have two scopes of equal aperture but differing f-ratios. If you have an extended object that can be encompassed by the FOV in both scopes, then the total number of photons captured of that object by each scope is the same. That is the 'myth'.
The object will initially look brighter in the lower f-ratio scope, but only because it's putting the photons into a smaller area. The number of photons from the object is the same. Equally size the object from both images and they will be equally bright because the number of photons captured is the same.
The low f-ratio scope will gather more total photons, but that is only because it has a larger FOV (and the attendant loss of resolution).
Why is the number of photons the same for this object? Because the strength of the signal is dependent on aperture and time, not f-ratio. If you want to capture an object faster, you need to increase aperture, not f-ratio. I don't know of any expert who disputes this basic law of physics.
I do think that people can quibble about the contribution of read noise to the S/N in the dimmer sections of an image taken with the higher f-ratio scope, but the point is that the total signal from the object is the same for both scopes.
dan
|
Galaxyhunter
Pooh-Bah
Reged: 01/02/06
Posts: 1251
|
|
I guess that I'm really dumber that a box of rocks. I just don't understand how the laws of physics for a photon can change between a point source and an extended object. I started this Thread a few weeks ago. And it was explained to me that an extended objects source is spread out over multiple pixels while a point source is on a single pixel. Now I haven't used many different scope/camera combinations, but in everything that I have use, even a single star occupies more that a single pixel. With that in mind, your pixels have absolutely ZERO clue where that photon of light came from that is striking it, whether it came from a faint wisp of a Nebula or a 18 mag star.
So basically, I would like the following statment explained why the SNR if different when the pixels don't know the source of the photon.
Quote:
For extended objects, the adu in each pixel goes as the inverse square of the f/ratio, independent of aperture. As a result, assuming shot noise, the SNR goes as the inverse of the f/ratio - independent of aperture.
With smaller aperture, same detector, and same f/ratio, you lose resolution, but the per-pixel SNR is the same.
For stars, the SNR expressions are much more complicated and are largely determined by aperture - but not entirely.
--------------------
Carl
My lousy skies at Hawkeye Observatory
|
Arkalius
scholastic sledgehammer
Reged: 08/03/06
Posts: 878
Loc: Orange County, CA, USA
|
|
Light that comes from a specific direction that enters the aperture of your scope will not all be focussed into a single point. It will be spread out over an area in a parttern known as the diffraction pattern, or sometimes called the Airy pattern (nothing to do with air, named after George Biddel Airy who was the first person to describe it). Most of the energy will be focussed in a center disc called the Airy disc. Concentric rings of light radiate out from there but these contain a very small portion of the light. The size of the Airy disc of a point source is proportional to the diameter of the aperture.
Technically, the apparant size of stars is so small that no combination of amateur telescope and camera could ever magnify one to be larger than a single pixel. However, because of diffraction as described above, the light from a star will spread out to a minimum of the size of an Airy disc which is typically larger than a pixel in most cameras, but not always. Seeing will always enlarge the appearance of stars in a fairly regular way, so that will also increase the size of a star image on a sensor. Also, tracking errors too small for an autoguider to bother with will have a small effect too.
We say that star's brightness is not affected by focal length because of the fact that no matter how much you magnify it, it's still a point source. All of its light will still go into one pixel. This is technically not true though for reasons stated above. Increasing focal length will magnify the Airy disc more and also exacerbate issues with seeing. So, while the star itself isn't magnified to be larger than a pixel, the blob of light it turns into will be, so they do spread out a bit. Generally though they remain confined to a few pixels and are pretty bright to begin with relative to the other things we photograph, and magnifying them doesn't spread them out a whole lot so their SNR doesn't really change much with respect to focal length in most cases.
Someone else already described what happens with extended sources. Keeping aperture constant and increasing focal length spreads the same light from an object out over a larger area. Assuming the same camera between both scopes, the pixels in the longer focal length scope will each receive a smaller chunk of the photon pie, and since they detect fewer photons, the SNR will be lower in this case. You do of course gain more resolution, but because of seeing and diffraction limits, at some point you are only resolving the blobs of the Airy pattern more rather than actual details in your target. This is oversampling. When you do this you gain nothing at the expense of SNR.
Undersampling will lose image resolution but the trade-off is that you get a higher SNR. So, undersampling is less of a problem than oversampling is.
--------------------
-Arkalius
11" Celestron SCT on Orion Atlas EQ-G
Celestron 100ED Refractor
8" Zhumell Dobsonian Reflector
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Quote:
I think that you've missed the point. The only thing that Stan is saying is that suppose that you have two scopes of equal aperture but differing f-ratios. If you have an extended object that can be encompassed by the FOV in both scopes, then the total number of photons captured of that object by each scope is the same. That is the 'myth'.
Hi-
I'm aware that is what he is saying, and I agree it is true to some extent - but I don't agree that it is particularly relevant - and I don't agree that it is really tied to film vs. ccd.
f/ratio is very fundamental and always has been.
With hyperstar at f/1.8 you can bounce around the sky taking nice image after nice image of objects with short exposure, while at f/10 you would have to sit on the same objects for a long time to get deep images.
It's pretty simple - the S/N for extended objects PER PIXEL depends on f/ratio. The S/N for extended objects PER SQUARE ARC-SECOND depends on aperture.
The reason per-pixel is most relevant to imagers is that they typically have a fixed ccd and a choice of ota's AND a choice of objects. You can see nice images of m101 that fill the field of the image, or wide shots that have m101 small in the field. People typically DON'T crop and blow up to match image scales - so it is per-pixel S/N that makes a fast f/ratio image deeper than slow f/ratio - independent of aperture. Even if you do blow up a fast image of the same aperture and same exposure, it will show less read and shot noise but reduced resolution.
To make it more concrete - if you think aperture is all that matters for practical imaging, then go ahead and take an image of m20 with a barlow at f/20. Why not 3 barlows? It's the same aperture and f/ratio doesn't matter??
If you keep the ccd the same and do not crop or blow up the image, which is what imagers tend to do - showing more of a wide scene when it is available - f/ratio will determine your exposure time independent of aperture. If you do crop and blow up - fast f/ratio will still show less read noise and shot noise, but reduced resolution.
Again - I'm still looking for actual physicists, optical scientists, or textbooks that take the perspective that f/ratio is a myth. I already mentioned Schroeder as supporting my view.
Frank
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Quote:
And it was explained to me that an extended objects source is spread out over multiple pixels while a point source is on a single pixel.
Point sources (stars) are much more complicated. It is because they don't scale simply with focal length. They aren't points, but a mixture of diffraction (the Airy pattern) and seeing. As a result, their size on a ccd depends on a mixture of f/ratio and focal length. For the HST in space, star sizes on the ccd are, again, only determined by f/ratio because they are diffraction patterns. For ground based telescopes, they are spread out by seeing and don't behave so simply.
So - they are much more complicated to describe in terms of SNR, f/ratio, and aperture. Some expressions for their SNR can be found in Schroeder's book on astronomical optics, where he contrasts them with extended objects.
Frank
|
HunterofPhotons
sage
Reged: 04/26/08
Posts: 266
Loc: Rhode Island, USA
|
|
Frank, my comments are below:
Quote:
Quote:
I think that you've missed the point. The only thing that Stan is saying is that suppose that you have two scopes of equal aperture but differing f-ratios. If you have an extended object that can be encompassed by the FOV in both scopes, then the total number of photons captured of that object by each scope is the same. That is the 'myth'.
I'm aware that is what he is saying, and I agree it is true to some extent - but I don't agree that it is particularly relevant ....Frank
*** That what is relevant? That the signal received of an extended object is the same for differing f-ratio scopes? It's all about the S/N, and the point of the myth is that the signal is the same.
Quote:
f/ratio is very fundamental and always has been.
You're just falling prey to Stan's provocative use of the word 'myth' to describe f-ratio. No one is saying that f-ratio is not a valid concept or descriptor other than you.
Quote:
With hyperstar at f/1.8 you can bounce around the sky taking nice image after nice image of objects with short exposure, while at f/10 you would have to sit on the same objects for a long time to get deep images.
*** No, the point is that if you shrink the f/10 image down to the tiny size of the f/1.8 the 'brightness' (total signal) of your target object is the same.
Quote:
It's pretty simple - the S/N for extended objects PER PIXEL depends on f/ratio.
*** Yes, no one disputes that.
Quote:
The S/N for extended objects PER SQUARE ARC-SECOND depends on aperture.
*** Yes, that's the point. If you want to increase the S/N of your object increase the aperture (or increase the time and/or decrease the noise).
Quote:
..... Even if you do blow up a fast image of the same aperture and same exposure, it will show less read and shot noise but reduced resolution.
Certainly it will have reduced resolution. As for the noise contribution, here's what Stan says about that:
"Ray Gralak \(Yahoo Groups\)" <yahoo@...> wrote:
> ... "object S/N" is substantially affected by f/ratio...
> the *minimum* exposure time to substantially overcome
> readout/dark/ sky noise is directly proportional to f-ratio.
Yes.
That dynamic applies to the minimum sub-exp time necessary to take a "sky limited" exp. But for *total* exp time, the f-ratio effect on object S/N is negligible (assuming near sky-limited subs). F-ratio &/or pixel size noise effects can be very significant for overly short exps and that's why it is advantageous to bin focus-mode and guide exps. It's also worth noting that this effect is usually limited to dim objects near the image's limiting mag and has little or no effect on the brighter objects, which have enough Poisson noise to overcome camera noise.
Although it may seem a bit counter-intuitive at first, the concept of "object S/N" is simple, elegant, and has great explanatory power. But most "pretty picture" enthusiasts don't much care about factual content (e.g. limiting mag) and are primarily concerned with the visual impact of the presentation image, which in large part can be characterized by "pixel S/N". Pixel S/N is strongly driven by f-ratio/pixel- size, which is essentially the same as the familiar "photographic exposure" f-ratio/exp- time dynamic. Consider the theoretical case of the "square nebula": a square nebula with no internal features that emits light equally over its full extent. If that nebula is imaged with a certain camera and aperture at two different f-ratios (i.e. focal lengths) then the resulting images will each capture the same number of photons and will equally represent the nebula's flux per square degree. But the un-resized images will exhibit different pixel S/N: the longer FL (slower f-ratio) image will show a larger nebula with lower pixel S/N, which can appear "noisier" than the shorter FL (faster f-ratio) even though the factual noise (per square degree) is no different.
Things start to get complicated when you exchange that "square nebula" for a real nebula that contains features of many extents and contrasts. The concept of S/N becomes insufficient to characterize these complications and a more accurate assessment requires MTF analysis, which is difficult and beyond the limitations of a forum discussion.
A simplification of the problem is to consider that S/N is associated with the ability to detect contrast levels. This association is primarily object related and "pixel S/N" alone is insufficient to ascertain contrast thresholds. In other words, changing the image scale (alone) does not have an appreciable effect on one's ability to detect features (apart from potentially destroying information via undersampling) . So even though a fast f-ratio image has higher pixel S/N than does an otherwise equal slow f-ratio image, it does not reveal any more factual information (unless camera noise is allowed to interfere; and if the faster image is undersampled then it will contain less information than the slow image).
So the bottom line is that a given nebula or galaxy will produce a gross average photon flux that could be considered as a normal photographic object, subject to the "f-ratio/exp- time" dynamic. Thus reducing an f/10 scope to f/5 allows the user to reduce the total exp time by 4x and still produce an image that superficially "looks" just as "bright". But that reduced image will contain much less factual information and will be objectively inferior (e.g. worse limiting mag, fewer and less visible details/features) .
Stan
Quote:
Again - I'm still looking for actual physicists, optical scientists, or textbooks that take the perspective that f/ratio is a myth. I already mentioned Schroeder as supporting my view. Frank
*** Did you post a link or quote Schroeder somewhere, because I don't know what you're referring to? Here's what Rolando has to say on this subject:
"Note this, however, that faster focal ratios do not mean more signal to noise in the age of digital imaging. The focal ratio does not determine the amount of information or the ability to reach deep into the sky for elusive faint detail. Only aperture does this, so whether you have a 5" aperture F5 or a 5" aperture F10, the signal to noise will be identical. The only difference is the field covered on the same size CCD chip.
Rolando"
|
Galaxyhunter
Pooh-Bah
Reged: 01/02/06
Posts: 1251
|
|
Quote:
Light that comes from a specific direction that enters the aperture of your scope will not all be focussed into a single point. It will be spread out over an area in a parttern known as the diffraction pattern, or sometimes called the Airy pattern (nothing to do with air, named after George Biddel Airy who was the first person to describe it). Most of the energy will be focussed in a center disc called the Airy disc. Concentric rings of light radiate out from there but these contain a very small portion of the light. The size of the Airy disc of a point source is proportional to the diameter of the aperture.
I'm in no way, shape, or form disputing this fact. What I don't seem to understand is, because of the Airy Disk, and the fact that is is now spread out over multiple pixels, doesn't that technically make it an "extended object". I mean the instant that the photons strike the CCD detector, it is what it is (spread out).
Another question: Do ALL photons create a "Airy" disk? If the photons from a 15th mag star create an Airy disk, then shouldn't the photons from a 10th mag Nebula do the same? I do understand the light from a star is basically a pinpoint, But I don't think M57 is a 1.5' wide blanket photon. Every pinpoint spot of the Nebula is crating photons, and as such, wouldn't those photons have to follow the same Airy disks rules?
--------------------
Carl
My lousy skies at Hawkeye Observatory
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
You have many interspersed quotes - but all from from self-taught amateurs. I still haven't seen a reference to a textbook S/N expression or a journal article. Schroeder's book is Astronomical Optics and he goes in some detail on SNR calculations for point and extended sources.
As long as you agree that pixel s/n is determined by f/ratio (independent of aperture) while square-arc-second s/n is determined by aperture (independent of f/ratio) then we do agree on that. But the problem with referring to "f/ratio as a myth" is that people think an 11" f/10 will make just as good images of nebulae in 5 minutes as an 11" f/1.8. That is just not true - if you put the images side by side, the 1.8 will look much better because it is faster - and the light is concentrated on a smaller part of the detector.
I think the relevant thing for aesthetic images is pixels - precisely because when you have a 1000x1000 pixel detector - you tend to look at the image at about the same size regardless of the OTA used to image. With a small f/5 lens and 5m exposure of m101, making a small galaxy, or a large f/5 reflector - you will have a little galaxy in one and a big one in the other - and they will both look similarly imaged because the pixel s/n is the same. The reflector image will be just as "deep" but show more details.
Again - if in practical terms aperture is all that matters, then adding 3 barlows to an f/10 telescope will be just as effective at imaging nebulae as hyperstar at f/1.8. The images will look fine side-by-side. Does anyone believe that? On the other hand, because f/ratio is so fundamental, a 5m image with a tiny fisheye at f/5 will show the same pixel values as a giant f/5 reflector in the same exposure. That is fundamental and important for people to know - and I wouldn't want that to be lost in all this myth talk.
Frank
|
Doubleglaze
sage
Reged: 11/01/07
Posts: 201
Loc: Pacific NW
|
|
This topic or threads very much like them have come up a number of times over the past year and the discussion and explanations have always helped to frame and solidify my understanding of the general topic. Thanks all for the information and debate, each new piece of information or explanation has triggered, at least for me, new ways of looking at the basic interactions between telescope and camera and the principles behind capturing satisfying images.
As this is a beginners forum I think it helps to take a step back from time to time and think about the fundamental points being discussed. Although I've been observing for more than 25 years, have studied physics at a graduate level, and design laser diode, detector, and miniature optical packages for a living, I'm pretty much a beginner in regards to astro-imaging. Even though what's being discussed here is really pretty basic it's taken a fair amount of thought and study to pull apart the details, and not to be snooty but if I've had a hard time grasping some of it then a true beginner will likely get lost in the weeds very quickly, especially when blanket statements like "aperture doesn't matter", " and "you only need to compare f-ratios" and "the f-ratio myth " tend to muddy the waters, at least to my thinking.
I think it's very useful for beginners to understand the basics as they ponder how best to spend their time and dollars in this hobby. So with that in mind I offer these thoughts, please correct me if I got something wrong. Also try and remember that there are probably 100s of people wondering "what the heck did that mean" and from time to time it may be useful to step back and redefine what is being discussed.
Cheers
Mark
--------------------------------------------------------------------------
The whole debate is really contingent on what you want to do with your telescope and imaging setup. There isn't one answer to what's the best aperture, image scale, focal ratio, CCD pixel size, etc. It's all determined by - what do you want to do, and how much do you want to spend?
For the vast majority of the people who stop by to chat on Cloudy Nights, astronomy is a hobby and we do it for fun. If it makes you happy, do your own thing regardless of whether or not its "right". As you learn more, you can do more.
Taking images of the night sky is pretty awesome even if you don't understand what you're doing. Just taking any image at all is pretty close to a miracle for someone from 200 years ago. Think of where we'd be scientifically if Galileo or Herschel could have called up SBIG and gotten a camera for, I don't know, 2 goats?
There are a couple of really basic things to get a handle on - what does aperture, focal length, focal ratio, tracking, image scale, CCD pixel size, signal, noise, and signal to noise ratio have to do with successful imaging. There's a ton more to learn about to take successful images, but I think these topics encapsulate most of the points being discussed in this thread. Of these topics, at least to me, SNR is the hardest to get a handle on. After than I'd say understanding resolution and image scale gets you a long way towards understanding how it fits together.
Things that don't make sense (at least to me), and are worth debating and or debunking (at least to your own satisfaction). When you stumble on these topics, you sometimes have to add in some missing details to understand the context -
You can't say "f-ratio is all that matters" without taking other things into account. If this were true than my 50mm f1.4 Canon lens would be among one of the best imaging scopes in the world. For example the Keck telescope on Mauna kea has an f-ratio of 1.75. Hmm, my Canon lens is a bit better than one of the largest scopes in the world? How can it be that a consumer camera lens is better than a world class observatory with a telescope that has an aperture equivalent to 10meters? http://www.spacecraftkits.com/KFacts.html
What I didn't say when I say f-ratio is all that matters is that I also have to describe the resolution and image scale that I'm able to achieve. Image scale depends on CCD pixel size and telescope focal length. Resolution depends on telescope aperture. Notice I need to define both aperture and focal length. One piece of information isn't enough to describe the optical and imaging system, you need several to paint a clear picture, so to speak. Clearly the Keck telescope and my Canon lens are different beasts, as evidenced by the differences in resolution - the Keck being able to see small things with more clarity, and image scale - my Canon having a much wider field of view, great for constellation photos, not so good for planetary nebula, although I did capture 6 pixels worth of M57 last week.
You also can't say "aperture is all that matters" or you might decide, as mentioned in this thread that adding a 2X, 3X, 5X Barlow to the optical system shouldn't change your imaging characteristics. Adding a barlow changes the focal length, which changes the image scale. Great for pictures of Saturn, not so good for imaging extended objects in any reasonable time frame (though I guess you could build a mosaic of M31 bit by bit - good luck on that). Changing the focal length changes the focal ratio, which changes the speed at which images can be captured because the incoming light gets expanded out (when making the focal length longer) over more pixels, making the signal per pixel lower and hence the SNR more poor, which drives you to need to make the exposure longer to recover some of the SNR.
More aperture = more light collection area = more photons per second delivered to the imaging system. More light = more signal. What you do with that light (i.e. what camera you use) determines image scale, SNR per pixel, and SNR per image. In general I'd claim you want the largest aperture you can get while maintaining the image scale you desire, which is another way of saying focal ratio and pixel size need to be considered along with aperture to have a meaningful discussion.
Signal to noise ratio implies a measurement, or more precisely the variation in a measurement, and hence an astronomical object doesn't have an intrinsic SNR. If a star emits photons in a forest, and no one records it, does it have a SNR?
All photons are treated by the optical system in the same manner regardless of where they came from - whether a point source or an extended object (ignoring optical aberrations and on-vs off axis for example). To nearly as many decimal places as you'd care to count all astronomical objects are infinitely far away as compared to the aperture of any amateur telescope. The impact of imaging through air means that all sources of light are spread out to some degree - there are really no point sources and no perfect Airy discs. Diffractive effects by having a finite aperture means that all light sources will be spread out by the optical system. For a single point source of light the Airy disc is a convenient way to describe what happens to the object's light as its diffracted by the optics and is expressed in terms of resolvability. For anything being imaged in a real telescope system you really need to look at the Modulation Transfer Function to understand what the optical system is doing to the spatial information (i.e. the night sky). Now that I think about it, if the apparent size of a point source (i.e. a single star) as imaged through the atmosphere is smaller than the resolution limit for the optical system, then the spread-out point source still be could be considered as a point source, where the point spread function is a simple Airy disc - think this is right but it's been a while so I could be wrong.
--------------------
Vixen VMC260L / Sphinx SXD
Pentax 75 SDHF
Canon 40D / 50mm f1.4 / 100mm f2.8 macro / 28-135 f/3.5-5.6 / 200mm f2.8L
http://www.astrophotogallery.org/showgallery.php/cat/500/ppuser/166
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Quote:
a true beginner will likely get lost in the weeds very quickly, especially when blanket statements like "aperture doesn't matter", " and "you only need to compare f-ratios"
Hi-
I'm not sure if you are referring to me with regard to such blanket statements - but I want to be clear that I never make such statements outside of a very specific context, e.g.:
For SNR on a per-pixel basis, of extended objects, f/ratio is all that matters. For SNR on a per-square-arc-second basis, aperture is all that matters.
These aren't blanket statements - they are precisely stated in a specific context - and the fact that both say very different things but both apply in practical ways is why f/ratio is so confusing.
In practical terms, beginners striving to fill pixels with photons will directly appreciate the benefit of imaging at fast f/ratio - whether with a large or small aperture scope.
Here is an example I happen to have. I didn't intend for it to demonstrate f/ratio, but it does.
Pacman at f/10 and f/6 is a web page showing my close-up of the central pac-man at f/10 and f/6. These are narrow-band images, so they are starved for photons and need long exposure. The f/10 image is 14x15m exposures, while the f/6 image is 6x15m. Just based on f/ratio, the values at each pixel, for nebulosity, would be about 3 times as high in each 15m exposure for the f/6 image. Since the total exposure is only about twice for the f/10 image, the f/6 image looks less noisy and cleaner. If you look carefully, though, the f/10 image has slightly more detail - but it is at the limit of seeing and guiding.
The key thing is - I am showing each image at the same pixel scale - which I believe is what imagers tend to do. The f/6 image covers more space - so it would make no sense to crop it and match the f/10 image.
There are myths and truths associated with f/ratio, and I do see people just dismissing f/ratio as irrelevant and a dated concept with ccd. I don't find that thinking to be beneficial, and I don't find it in any textbooks or journal articles.
With smaller aperture and the same f/ratio, the pac-man image would be just as saturated on the pixels, but the image would have lower resolution since it is smaller.
Frank
|
HunterofPhotons
sage
Reged: 04/26/08
Posts: 266
Loc: Rhode Island, USA
|
|
Quote:
You have many interspersed quotes - but all from from self-taught amateurs. I still haven't seen a reference to a textbook S/N expression or a journal article. Schroeder's book is Astronomical Optics and he goes in some detail on SNR calculations for point and extended sources......
Frank
It's amusing that you dismiss Roland's take on the f-ratio myth as coming from a "self-taught amateur".
dan
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Actually I don't know which of several Rolands it refers to. I was replying to someone's comment about physicists/scientists dismissing the f/ratio, so in that context I am interested in people who have advanced degrees in a relevant field and are publishing papers in peer reviewed journals on the topic. If the Roland you are referring to has done so on this matter, I'm happy to stand corrected and take a look.
Frank
|
HunterofPhotons
sage
Reged: 04/26/08
Posts: 266
Loc: Rhode Island, USA
|
|
Quote:
Actually I don't know which of several Rolands it refers to. I was replying to someone's comment about physicists/scientists dismissing the f/ratio, so in that context I am interested in people who have advanced degrees in a relevant field and are publishing papers in peer reviewed journals on the topic. If the Roland you are referring to has done so on this matter, I'm happy to stand corrected and take a look.
Frank
There are, indeed, several Rolands out there, but there is only one Rolando (dancer extraordinaire). I don't believe that he has an advanced degree, nor that he publishes papers in a peer reviewed journal. I guess then, that, a priori, his a posteriori knowledge would be suspect. <g>
Rolando's Images and Essays
His Business
dan
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Yes - I stand by my story. I am looking for peer reviewed publications and/or textbooks by a scientist/engineer with a background in imaging theory, detector theory, etc. Such a person would be correspondingly trained in specialized graduate level courses and do original, peer reviewed and published research. In other words - a scientist/physicist, as mentioned in the earlier post. A text I happen to cite is, again, Astronomical Optics by Schroeder. Since the claims about f/ratio are not specific to amateur work, I would like to it similarly spelled out outside of the web/amateur realm.
I am happy to agree, disagree, discuss, etc. with anyone on this subject. Meanwhile, I post examples of my own to demonstrate my points, and I cite appropriate texts. I know there are recent amateur web pages that disagree with me, but I don't know what references and formal training in optics/imaging they use as a basis.
I don't know why it is so hard to separate pixel SNR from angular resolution... Stack 3 barlows and see what you get in 5 minutes. Great resolution - terrible pixel noise - and a tiny, useless image that looks awful. Pixel snr for nebulosity depends on f/ratio. Beginners and advanced imagers alike want lots of photons in each pixel to beat read and shot noise - f/ratio will do that for you. For nebulosity. With resolution determined by aperture.
Frank
|
Kolenka
Pooh-Bah
Reged: 06/01/08
Posts: 1009
Loc: Seattle Area, WA, USA
|
|
Quote:
I don't know why it is so hard to separate pixel SNR from angular resolution... Stack 3 barlows and see what you get in 5 minutes. Great resolution - terrible pixel noise - and a tiny, useless image that looks awful. Pixel snr for nebulosity depends on f/ratio. Beginners and advanced imagers alike want lots of photons in each pixel to beat read and shot noise - f/ratio will do that for you. For nebulosity. With resolution determined by aperture.
Frank
I actually wouldn't think that shot noise really makes a lot of sense on a per pixel basis. I know it seems intuitive to consider shot noise this way, but there is a lot about quantum mechanics that operates in a counter-intuitive fashion. The classic example being the double slit experiment, and how the distribution would occur the same way, regardless of how the photons arrived at the slit.
It stands to reason that there is also the possibility that shot noise deviation is predominately determined by aperture and time (and thus total photon count from the object) rather than the photon count hitting a single pixel. The more total photons collected from the object, the lower the shot noise, even if you spread the light out. Because shot noise doesn't exist in a vacuum though, you say 'longer focal length makes the image dimmer, and I see more noise' and be completely right. But we aren't guaranteed what levels of shot noise and read noise we are seeing just by looking at it. One might be able to do some experiments to find out though.
This is one thing I'd love to see clarification on from a reliable source.
--------------------
Orion XX12 / Orion 80ED OTA / AT66ED
Nagler 7T6, 9T6, 13T6, 17T4, 26T5
Canon XS, TIS DMK 31AF03, AstroTrac TT320X
Northwest Astro Photoblog
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
I don't think there is any subtlety regarding quantum effects here - any more than there is when using an old brownie camera and 120 film, or the latest SBIG ccd.
The equations regarding f/ratio are based on simple geometrical optics and can be found in any related text. More advanced texts will relate them to the concept of etendu and the Lagrange invariant of an optical system. They may then relate those, and the fundamental properties of f/ratio, to the second law of thermodynamics. This is all independent of quantum stuff.
Read noise is a detector issue and it applies for the most part on a per-pixel basis.
Shot noise goes as the square root of the number of counts, and here a 'count' is the number of photons arriving at a pixel. You only have counts on a per pixel basis. If you have uniform flux, then the mean count at each pixel will be the same, related to f/number, and the noise will be at the same level, but independent for each pixel.
There are several other noise sources in a ccd.
When you then look at the image on the screen, there will be noise due to the variation in pixel levels from the noise sources. Increasing the signal counts per pixel, via smaller f/number, will overcome the noise sources in each pixel and in the shot noise of the signal photons.
There are many books on this and related topics - it's a matter of what level you would like to see. Schroeder is more graduate level I guess. There are many books on statistical optics - I like Frieden, Probability, Statistical Optics, and Data Testing.
There are plenty of references consistent with my understanding of things - I'm looking for one on the f/ratio=myth side, which suggests that "aperture is all that matters and you gain nothing with faster optics."
Some scope manufactures, e.g. Takahashi (at f/2.8!), go to great effort to make fast astrographs. Ceravalo offers one that switches between fast and slow. Why bother? To trade off resolution with depth and field of view. And all telescopes, from the Hubble to amateur ota's, are still, and properly, described by f/ratio.
The ceravalo astrograph is described as offering "the optically fast f/4.9 wide field mode, and the f/9 high resolution mode." Optically fast. Even today, with ccd's, it still plays a role.
Ceravalo astrograph
Here is the Takahashi quotation:
"This fast field photo visual astrograph is perfect for the new mega pixel digital 35mm cameras and CCD cameras. The great speed will enable the imager to record faint nebulosity invisible in slower instruments. "
Takahashi Epsilon astrograph
Frank
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Here is a follow-up quotation from the astro-physics web site:
"The 160EDF operating at f5.7 (0.76x compression) is an ideal match for the large format STL series cameras like the STL6303 and STL11000. Extreme detail can be captured in very short time thanks to the fast focal ratio. This accessory is a must for H-alpha imaging where speed and resolution allows one to achieve incredible results."
I'm not sure who wrote this and if everyone at the company agrees with it - but I sure do. Especially for Ha imaging, fast f/ratio is a big win, and I would recommend that accessory for a relatively slow refractor.
The text is from the 160 f5.7 telecompressor corrector (160TCC).
astro-physics web site
Basically it reads just like a hyperstar or any other ad pushing the perfectly valid benefits of fast f/ratio.
Frank
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
IMO, reducers help not because you're altering the f/ratio (which they do) but because it reduces focal length thereby focusing more light on individual pixels.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
That is precisely what a fast f/ratio does. It puts more light from extended objects (nebulosity) into each pixel. And the amount of light (adu value) in each pixel is independent of aperture for the same f/ratio. Only the f/ratio matters in determining the value of the pixel. The aperture, for the same f/ratio, will determine how big the object is, and its level of detail.
Frank
|
Jared
Carpal Tunnel
Reged: 10/11/05
Posts: 2531
Loc: Piedmont, California, U.S.
|
|
Frank,
Would you agree with the following:
If you could choose a camera pixel size to match your focal length so that images were "optimally" sampled (whatever that means), then you would get identical results in terms of SNR no matter what focal ratio is chosen?
In other words, would an 8" f/2.8 scope used with a camera with 2.7um pixels yield identical results to an 8" f/10 scope used with a camera with 9.6um pixels? You should get the same field of view (assuming the total number of pixels is held constant), the same spatial resolution (1 arc second per pixel), and the same signal to noise ratio (assuming read noise and thermal noise were constant between both cameras). Do you agree with that?
Here is one more way of saying the same thing. If you match your camera and your telescope so that you are optimally sampling your subject, then you should get the same results no matter what the focal ratio is. The same exposure times would be required, and the same image quality would result no matter what the focal ratio happened to be.
Obviously, if you choose not to use a pixel size that is "optimal" for your scope and typical seeing conditions, then by undersampling you can shorten both integration time and subexposure time. This undersampling can be achieved by choosing larger pixels, by choosing a faster scope, by installing a focal reducer, or by in camera binning.
HyperStar is a perfect example of this. An 8" SCT shooting at f/2 (Hyperstar setup) is going to allow much shorter subexposures for shot noise to swamp read noise, and much lower total integration time for a given image SNR. It will also provide much wider fields of view. So why aren't we all using HyperStar setups or their equivalents? Because the improved SNR and shorter exposure times came at a cost... Spatial resolution. Odds are pretty good that an 8" f/2 telescope is undersampled. With 7.4 micron pixels (typical of many cameras) I'd only be getting 3.8 arc seconds per pixel. That might be great for some wide field shots, but clearly I would be giving away resolution vs. the same scope and camera at Cassegrain focus.
I think what Stan Moore was really trying to get across in his provocatively titled essay was not that focal ratio was irrelevant, but that it is possible to produce similar results no matter what the focal ratio involved. An f/12 telescope is not necessarily worse for deep sky imaging than an f/6 scope--you need to look at the entire imaging train. If you match your camera to your scope, there is actually a pretty wide range of focal ratios that will work. If your scope and camera are appropriately matched, then the only way to improve SNR is to either increase aperture or increase integration time.
--------------------
- Jared Willson
- TMB 152 f/8 Apochromat
- Fluorostar FLT-110 w/ TEC optics
- Stellarvue SV80S
- Astro-Physics Mach1 GTO
- Takahashi Teegul SP Mount
- STL-11000
|
Galaxyhunter
Pooh-Bah
Reged: 01/02/06
Posts: 1251
|
|
Terminology,
I think that people is confusing f/ratio as it pertains to photograph & astronomy. Most everybody knows that in astronomy,
f/ratio is the ratio of where the image comes into FOCUS. I.E.; 105mm f/6 = 630mm
16" f/4.5 = 72" Changing the f-ratio changes the focal LENGTH.
In photography, the f/ratio is the amount of light that is passed through the lens in relationship to the main objective lens.
I say this because, back in the day when I used to process my own film & then print them out, When I changed the aperture
stop on my enlarger, I never had to refocus. If the f/ratio on a camera lens is the same thing as on a telescope, then refocusing
my enlarger would have been mandatory when I changed the aperture stop. All I had to do when I switch from f/8 to f/16 is to
double the time I had to have the enlarger lamp on. Test this out yourself. Dig out your SLR (DSLR), mount up a fixed length lens.
Focus on something , then change the aperture stop. DID it all of a sudden fall completely out of focus?
Now I'm (and I don't think anybody else is) disputing the fact that a higher f/ratio scope requires longer exposure times. I don't
have a whole lot of technical knowledge on doing AP. But common sense tells me the imaging is a combination of BOTH
f/ratio AND aperture. The way I see it, the equation should look something like
Captured Photons = f/ratio * aperture * exposure length.
You can not change one part of the equation without effecting some other part.
If I'm all wet in this thinking, Somebody please tell me & I'll stand corrected.
Now everybody can (and probably will) argue until they are blue in the face about SNR, mainly because hardly anybody can agree
just what is the definition of SNR. I, myself tries to stay out of those.
--------------------
Carl
My lousy skies at Hawkeye Observatory
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
Quote:
Now I'm (and I don't think anybody else is) disputing the fact that a higher f/ratio scope requires longer exposure times. I don't
have a whole lot of technical knowledge on doing AP. But common sense tells me the imaging is a combination of BOTH
f/ratio AND aperture. The way I see it, the equation should look something like
Captured Photons = f/ratio * aperture * exposure length.
You can not change one part of the equation without effecting some other part.
If I'm all wet in this thinking, Somebody please tell me & I'll stand corrected.
But with CCDs there's a forth element to the equation - pixel size (OK, and QE and...).
What's important to understand is that the pro-FR argument maintains a fixed pixel size in which case the pro-aperture argument actually agrees - as long as the pixel size remains fixed.
What the pro-aperture argument says though is that if the image scale (or arc seconds of the sky per pixel) is fixed rather than the pixel size, then aperture and only aperture (OK, and QE and...) determines the number of photons per pixel collected. Sure, an 8" F/10 would require pixels twice the size of an 8" F/5 to gather the same number of photons, but then the image scale is the same in both cases.
However what the pro-aperture argument also says is that with a larger aperture it is possible to have both higher resolution and gather more photons per pixel with the same exposure time if you have appropriately sized pixels. On the other hand a smaller aperture scope can have either equal or greater resolution per pixel than a larger aperture scope or it can gather more photons per pixel (or it can have neither) but it can't do both.
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
Galaxyhunter
Pooh-Bah
Reged: 01/02/06
Posts: 1251
|
|
Quote:
But with CCDs there's a forth element to the equation - pixel size (OK, and QE and...).
Thank you for pointing this out Dean, so the modified equation might look like
Captured Photons = f/ratio(Focal length) * aperture * exposure length * Pixel size * Camera efficiency.
--------------------
Carl
My lousy skies at Hawkeye Observatory
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Hi-
I think it might make most sense if I argue from the other side - i.e. the myth side. Here goes.
It is a myth that an 11" f/5 with a given ccd will take as good an image in 5 minutes as an 11" f/10 will in 20 - even though the "f/ratio rule" says they should be the same.
I think this sums up, I guess, what the myth folks are saying - and - sure - I agree with that. No problem. But the thing is - I don't know anyone, in film or ccd, who would say something so broad - specifically that the images would be "the same."
What people do say is - with 11" f/5, you can go much deeper in less time than with 11" f/10, and you will capture more faint nebulosity. But the 11" f/5 will have less resolution and detail - in the bright regions anyway.
What I would further say is - the pixel values, for nebulosity, only depend on f/ratio, independent of aperture, for a given ccd. That is what I consider the f/ratio rule - and it is rock solid - for film and ccd.
Now - the myth folks then want to blow up the 11" f/5 image to the same scale as the 11" f/10. I say - that is not what people do. If you have f/5 and a nice, deep wide field - you don't blow it up and crop it. You leave it at the same pixel scale and have a smaller, deeper, and more contrasty image. You would not spend a ton on an f/2.8 astrograph only to blow it up and crop out the central region to match f/10 for some reason.
In summary - I don't think there ever really was an f/ratio myth. I don't see anything pertaining to film vs. ccd. Some people might be confused and think that 11" f/5 will magically work as well in all respects as f/10 in 1/4 the time - but they would be wrong. They would be right, though, in saying it goes deeper at the expense of resolution - but that is true of film and ccd.
I don't know if this helps - but I'm trying.
Frank
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
Quote:
Thank you for pointing this out Dean, so the modified equation might look like
Captured Photons = f/ratio(Focal length) * aperture * exposure length * Pixel size * Camera efficiency.
To calculate the actual number of photons, you also need the photon flux (photons per second per unit area) entering the scope.
Stan Moore defines it (sortof) as:
ObjectFlux *QE * Time * Aperture * ScopeEfficiency
where ObjectFlux is photons/second/square-meter and aperature is square-meters.
But it's late and I'm not sure what this solves to: photons per -- what? I'd have to slog through it and resolve the units and then figure out how pixel size fits in, and I'm just not up to it right now. I seem to recall going through this once before and found that it doesn't solve to anything useful unless you also plugged in FL (oops! aperture and focal length - that sounds like ... FR!).
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
gavinm
scholastic sledgehammer
Reged: 08/26/05
Posts: 804
Loc: Auckland New Zealand
|
|
The Stan Moore formula simplifies down to photons.
--------------------
Gavin
Mt Albert Grammar School Observatory
Auckland, New Zealand
http://www.mags.school.nz/astronomy/index.html
12" LX200R F6.8 AP
SBIG ST7-XME + CFW10
Moonlite SCT focuser w/ temp
Skywatcher Equinox ED80 Pro (ADM dovetail)
+ other stuff
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
Only focal length and detector/pixel size determine Image scale.
Aperture determines how many photons enter the optics.
That flux is divided amoung the pixels based on FL and pixel size.
The problem is that a change of f/ratio could be a change of FL or of Aperture, or both.
F/ratio alone doesn't tell us everything we need to know to determine what changes we'll see at our detector.
And if you want to actually test this, make an aperture mask for your scope, you've just changed the flux entering the scope, the f/ratio but kept a fixed FL and a fixed pixel size.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Quote:
F/ratio alone doesn't tell us everything we need to know to determine what changes we'll see at our detector.
f/ratio is ALL you need to know to compare the pixel value of nebulosity for a given detector with one ota or another. Whether for film or ccd. The fact that this point is being lost is what concerns me about this myth talk. I believe even the proponents of the f/ratio myth would agree to this (pixel value is constant for same camera and same f/ratio, regardless of aperture) - and what they differ on is the loss of information because the image size is reduced with smaller aperture for the same f/ratio.
But if you image a nebula for 5m with one ota of any aperture, and then you image it with the same detector and any other ota with the same f/ratio, the pixel value for the nebulosity will be the same - independent of aperture. I think it's important for this point not to be lost. Even beginners with experience in photography are familiar with this - and it is equally true in astrophotography.
f/ratio alone does NOT tell you what the image will look like and how sharp it will be. That is where aperture plays a role. But a slow lens is still a slow lens, regardless of aperture.
Frank
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
Quote:
The Stan Moore formula simplifies down to photons.
That's just it - at some point you have to consider area at the image plane - be it pixel size or CCD chip size.
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
Qkslvr
Pooh-Bah
Reged: 06/23/06
Posts: 1316
Loc: NE Ohio, US
|
|
Quote:
Quote:
F/ratio alone doesn't tell us everything we need to know to determine what changes we'll see at our detector.
f/ratio is ALL you need to know to compare the pixel value of nebulosity for a given detector with one ota or another.
Frank
I think I can agree to that.
--------------------
Mike
Onyx 80ED/N8/CG-5/40D
|
Dean
Post Laureate
Reged: 12/31/04
Posts: 4910
Loc: Bailey Co Elev 8780 feet
|
|
Quote:
f/ratio is ALL you need to know to compare the pixel value of nebulosity for a given detector with one ota or another. Whether for film or ccd. The fact that this point is being lost is what concerns me about this myth talk. I believe even the proponents of the f/ratio myth would agree to this (pixel value is constant for same camera and same f/ratio, regardless of aperture) - and what they differ on is the loss of information because the image size is reduced with smaller aperture for the same f/ratio.
With all due respect Frank, if you read what Jared and I have been saying, your point is not being lost, and in fact we both agree with it.
What we are getting at is that while this is true, it hinges on the requirement that the camera (or pixel size) remains fixed.
--------------------
"Don't sweat the petty things and don't pet the sweaty things" - George Carlin
deanrowe.net/astro
Whats with that avatar?
|
Jared
Carpal Tunnel
Reged: 10/11/05
Posts: 2531
Loc: Piedmont, California, U.S.
|
|
Quote:
<snip>
But if you image a nebula for 5m with one ota of any aperture, and then you image it with the same detector and any other ota with the same f/ratio, the pixel value for the nebulosity will be the same - independent of aperture. I think it's important for this point not to be lost. Even beginners with experience in photography are familiar with this - and it is equally true in astrophotography.<snip>
Yes, we all agree with this. Where we differ is the assumption that the detector must remain constant. In practice, most advanced astrophotographers choose a pixel size to match their focal length (or vise versa) and a chip size to match their budget. Pixel size is as much a variable as focal ratio and focal length. That is where the statement about the "myth" comes in. It's not that anyone argues your points--you are accurate. Where we disagree is on the assumption that the detector must be held constant. Within a certain range, you can choose a pixel size and chip size that matches whatever focal length and focal ratio you happen to have. If you do this, you can make that f/8 or f/12 scope every bit as good for imaging as an f/2.8 scope.
Am I saying an f/12 MCT is a good choice for wide field imaging? Of course not. There are practical limits on the available chip and pixel sizes. Also, there are limits on how large an unvignetted, flat field a given scope design and aperture can produce.
Still, there is an element of truth to the "focal ratio myth" and that element of truth is that it is possible to arrange things so that scopes of different focal ratios yield equivalent per pixel intensities and equivalent image SNR's for a given aperture and exposure duration. Pixel size is as much a variable as focal ratio.
--------------------
- Jared Willson
- TMB 152 f/8 Apochromat
- Fluorostar FLT-110 w/ TEC optics
- Stellarvue SV80S
- Astro-Physics Mach1 GTO
- Takahashi Teegul SP Mount
- STL-11000
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Quote:
With all due respect Frank, if you read what Jared and I have been saying, your point is not being lost, and in fact we both agree with it.
What we are getting at is that while this is true, it hinges on the requirement that the camera (or pixel size) remains fixed.
Ok - I'm just trying to state f/ratio properties in a fundamental way that people can agree with. I have the impression that not everyone on this thread is willing to agree with even that part - but I'm not sure.
So - to me - this rule about f/ratio and a fixed detector is the f/ratio rule, and it applies to film and to ccd and is fundamentally linked to the second law of thermodynamics. There is nothing mythical about it.
Separately, apparently some people have the impression that a fast lens of the same aperture will capture as much detail and depth in less time than a slow lens with the same detector. That is false, but I don't know how many people ever really thought that - since a smaller image will have less detail.
Third, some people compare a blown up and cropped fast image and compare it to a slow image and say there is no difference. I think that is misleading and misses the point that fast lenses win you wider field and fainter nebulosity for the same aperture, and the images should be viewed at the same pixel scale since that's how people tend to present astro-images.
Fourth - I wanted to point out that I don't know of any references outside the web/amateur realm that make the various "myth" claims about fast optics. I have seen diatribes written about how hyperstar ads are misleading since low f/ratio has no benefit (aperture is all that matters) - and I think that is a misunderstanding rooted in misconceptions promoted by amateurs.
Finally - on the issue of pixel size and such. This isn't unique to recent ccd work since the same trade-offs of sensitivity and resolution apply to film with different ISO (or ASA) - despite not having well defined pixels.
In order to characterize fully an optical system, you would need to know everything about it and the detector. But once you do that, if you replace the optics with something else of the same f/ratio, nebulosity will have the same pixel values. And if you replace the optics with a lower f/ratio, the pixel values will go up with the inverse square of the f/ratio and your SNR will increase - most notably in faint regions of nebulosity - independent of aperture.
Frank
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Quote:
Still, there is an element of truth to the "focal ratio myth" and that element of truth is that it is possible to arrange things so that scopes of different focal ratios yield equivalent per pixel intensities and equivalent image SNR's for a given aperture and exposure duration. Pixel size is as much a variable as focal ratio.
I guess that is yet another issue - but it goes beyond f/ratio. I think it is a separate issue because the main web page that discusses the f/ratio "myth" does a comparison between two different 'scopes with the same detector - and blows one image up to compare resolution. So - at least that version of the "myth" doesn't require different detectors to make its "point."
I guess I would call your thing the "optical system myth," which is that f/ratio tells you everything about an imaging system. Again - I'm not sure who would believe that.
In practice, imaging optics are designed to create a certain maximum spot size over a certain field size. If you use pixels much larger than the spot size, then your optics are over-designed for the detector. If your pixels are well matched to the spot size, but the detector is much smaller or larger than the specified field size, you also have a design mis-match.
But - regardless of what detector you choose, you can be sure that nebulosity will have higher pixel values for the fast lens than for the slow lens - independent of aperture. The resolution you get will depend on aperture and pixel size. If you use bigger pixels, they will collect more photons - but they will still collect even more with a faster lens - independent of aperture.
Plus - as I said in the previous note - the same issues of changing pixels apply to film of different speeds. In terms of film photography, the ISO is a variable also. That's why the "sunny 16" rule is based on ISO (or ASA) 100 film. On a sunny day with 100 film at 1/100 second, shoot at f/16. It's still true with ccd cameras - independent of aperture. But just like with film, you have to compensate if you use faster film than ISO 100.
Frank
|
Jimmy2K63
Pooh-Bah
Reged: 04/26/09
Posts: 1188
Loc: Kentucky
|
|
Quote:
Quote:
Same CCD, 4":F10 vs 8":F5. Both have a focal length of 40".
Using the same ccd (assume a small CCD less than 1" diagonal), the image "size" should be exactly the same. M51 would look the same in both images.(I know the resolution will be different as the airy disk is defined by the diameter of the scope)
Here is the question. What about the "brightness" of the image? Will the exposure times be the same?
For the galaxy itself - not the stars in the scene - you can ignore aperture and just go by f/ratio in terms of brightness or adu in each pixel. The f/5 will be four (2^2) times brighter than the f/10.
This would also be true if you used a 200" aperture f/10 - the 8" f/5 would have 4 times the adu for the extended glow from the galaxy.
Both film and ccd would be "brighter" with the 8" - but the galaxy would be much larger with the 200". No myths apply - this stuff is as old as the hills.
Frank
Here is my take on this question.
8" f/5 = 40" F.L.
4" f/10 = 40" F.L.
Both images would be the same size. This has nothing to do with the focal ratio and everything to do with the focal length.
Light Gathering:
Area of mirror/lens" = pi*radius^2
so we are left with pi*4^2 versus pi*2^2
The 8 inch has four times the light gathering area as the 4 inch, so the imaging time on the 8 inch could be reduced to about 1/4th of the exposure time of the 4 inch. That's assuming a linear relationship, i.e. no "reciprocity failure." In the real world, that's a real issue, the place where it gets sticky is in long exposure times and very short ones, but in general, I feel you would be safe using that formula.
--------------------
http://astronomyguy63.blogspot.com/
LXD75 SN6-UHTC
Cave Astrola 10" f/5
Garrett 15x70/FarSight
Canon XS (1000D)
|
freestar8n
sage
Reged: 10/12/07
Posts: 299
|
|
Quote:
The 8 inch has four times the light gathering area as the 4 inch, so the imaging time on the 8 inch could be reduced to about 1/4th of the exposure time of the 4 inch.
This is not going to surprise many people because you have doubled the aperture at the same focal length and the image will now be 4 times brighter. Where it gets interesting is when you go from 4" f/10 to 2" f/5. Even though you have cut the aperture in half, the image will be as bright as it is in the 8" f/5. That is because the image is much smaller. Your aperture area has been reduced by a factor of 4, but your image area has been reduced by a fact | | |
Previous
Index
Next
Threaded
Edit
Reply
Quote
[Re: