56 replies to this topic

[JWolf_21]

Hi!

 

After reading around for a while, I think I have found the answer to my noob question but not 100% sure.

 

I'm used to hear to some of my astrophotography friends, and many internet posts, say "the exposure time of this image was 9 hours.... 3 hours of Ha, 3 hours of OIII and 3 hours of NII" or whatever filters they are using, RGB, whatever...

 

From my point of view, the exposure time of the full picture should be 3 hours, not 9, since you are collecting data of different, not-overlapping, parts of the spectrum. The procedure took you 9 hours, of course (plus probably 3 more hours of mounting-unmounting the telescope that you're not counting), but you collected photons only for 3 hours in each band, so the exposure time of what you're showing to me is 3 hours.

 

It's like if I do regular photography with a DSLR of a beach landscape, I shoot for 1 second, and I say that the exposure time is 3 seconds because RGB...

 

I have tried to discuss this topic several times with them and I learnt to shut up

 

Any light on this? (pum well-intented)

Edited by JWolf_21, 19 April 2025 - 12:01 PM.

[smiller]

You're actually asking a fairly complex question once you include narrowband filters and you are trying to draw an equivalency to daytime one shot color cameras.    If "how long someone exposed a target" is dependent not on total time imaging a target but some other metric like "how many photons did I get", then it's quite complicated.

 

So instead of all that complexity, people just express total exposure time (often called "total integration time") as the total time the camera and telescope are exposing the target, regardless of whether that be RGB with a color camera, a mono capture with separate R, G, B color filters, a mono camera with Luminance, R, G, B, Ha, Siii, Oiii filters, a color camera with a dual narrowband Ha/Oiii filter, or even if all the time was shot with a mono camera with no filter.

 

What people do is describe the situation under which they exposed for "X hours" and it's up to the reader to interpret that appropriately.

 

For instance, if someone says "I exposed the target for 9 hours with a mono camera for a monochrome image", I know they got about 3x as many photons as if they said "I exposed the target for 9 hours with a one shot color camera".

Edited by smiller, 19 April 2025 - 12:51 PM.

[UP4014Fan]

Interesting point. One could argue that the integration time is three hours. But, what if, an AP elected to expose one band more than another to emphasize it or draw it out. Is the integration time the minimum, the maximum, or an average?

To use a terrestrial analogy, when I was in Prague last year, I shot two time exposures of the market. One was based on the church steeples, which weren’t well lit, and the other of the market square, which was. I combined the two in Photoshop when I got home. All of this was mimic the dynamic range of the human eye. So, was the exposure time 5 seconds or 30 seconds? Or 17.5?

Edited by UP4014Fan, 19 April 2025 - 12:12 PM.

[JWolf_21]

Yes, I set a fairly simple example, with round numbers, to avoid the complication of different exposure times per band... but I see we all agree that the total integration time is "meaningless" in the sense that one needs to know the details of each channel to really know the exposure time of each part of the spectrum.

[gsaramet]

Just think you are shooting LRGB - say 10 hours L, 4h R, 3h G and 5h blue. What's your integration time?

 

For narrowband, one might shoot 20h Ha, 50h Oiii and 30 Sii. Again, what's your integration time?

 

 

As smiller said, the discution can get pretty complicated. 

 

However, it's also pretty simple: how much time did you spend capturing light? 

 

Note: why would one shoot different times? in order to get similar noise levels on different filters. 

[JWolf_21]

 

However, it's also pretty simple: how much time did you spend capturing light? 

 

I'd say that's irrelevant for the picture itself... that's how long took you to perform the procedure, but doesn't say anything about how much time did you collect each wavelength. That's why I think the total integration time is useless when describing an image.

 

For everything else, I agree
 

[rgsalinger]

You're actually asking a fairly complex question once you include narrowband filters and you are trying to draw an equivalency to daytime one shot color cameras.    If "how long someone exposed a target" is dependent not on total time imaging a target but some other metric like "how many photons did I get", then it's quite complicated.

 

So instead of all that complexity, people just express total exposure time (often called "total integration time") as the total time the camera and telescope are exposing the target, regardless of whether that be RGB with a color camera, a mono capture with separate R, G, B color filters, a mono camera with Luminance, R, G, B, Ha, Siii, Oiii filters, a color camera with a dual narrowband Ha/Oiii filter, or even if all the time was shot with a mono camera with no filter.

 

What people do is describe the situation under which they exposed for "X hours" and it's up to the reader to interpret that appropriately.

 

For instance, if someone says "I exposed the target for 9 hours with a mono camera for a monochrome image", I know they got about 3x as many photons as if they said "I exposed the target for 9 hours with a one shot color camera".

Aren't you assuming that the OSC camera's bayer matrix is reducing the number of photons being translated into charge by a factor of 3? I would think that the total loss in QE is much less than that, maybe 30 percent or so. I agree that it's just a term of art to use the total clock time as a general rule when describing the data collected in an imaging run. 

[JWolf_21]

Aren't you assuming that the OSC camera's bayer matrix is reducing the number of photons being translated into charge by a factor of 3? I would think that the total loss in QE is much less than that, maybe 30 percent or so. I agree that it's just a term of art to use the total clock time as a general rule when describing the data collected in an imaging run. 

Because of this, I set a simple mental experiment just with a monochrome camera (besides my DSLR analogy, which maybe was not the best example)... to avoid more complex situations and just understand the basics. I think monochrome and OSC exposure times cannot be directly compared due to Bayer matrix (unless dither and drizzle are performed, as stated in another parallel different topic of this forum), but this is a different story...

Edited by JWolf_21, 19 April 2025 - 01:26 PM.

[smiller]

Aren't you assuming that the OSC camera's bayer matrix is reducing the number of photons being translated into charge by a factor of 3? 

 

Yes, but I was speaking in very rough terms, disregarding the spectral distribution of the target, the human eye's higher sensitivity to green, the red blocking filter on land cameras, and the spectral absorption characteristics of the bayer filter dyes.

 

Based on the published spectral characteristics, I have actually calculated the integration of the total photon capture of a few common OSC astro sensors compared to the mono equivalent across the spectrum assuming a flat spectral target and you do capture only about 1/3 the amount of light, but some OSC cameras have enough overlap in the absorption of the various R,G, B channels that it can as high as 50%.   But that is a 2x reduction at best, much more than just a 30% reduction.  Certainly OSC cameras cut more than 30% of the photons reaching the sensor compared to a mono version of the same sensor.

 

"1/3" was the simple answer, but just like anything, the perfect answer is infinitely more complex than the simple answer.  

Edited by smiller, 19 April 2025 - 01:52 PM.

[gsaramet]

I'd say that's irrelevant for the picture itself... that's how long took you to perform the procedure, but doesn't say anything about how much time did you collect each wavelength. That's why I think the total integration time is useless when describing an image.

 

For everything else, I agree
 

It can easily get more complicated Consider a 5x5 = 25 panel mosaic with 20% overlap and a 10 hour integration per panel. Let it be a color camera.  What is the integration time?

 

Now extrapolate to SHO image with different times per channel AND RGB for star colors

 

And this is just math tricks. The REAL McCoy is this: 

 

One hour of a channel is NOT equal to one hour of the same channel. An hour of exposure at 15 degrees altitude is not the same as an hour of exposure at zenith. An hour of exposure in Bortle 8 is not equal to an hour of exposure in Bortle 3. An hour of Oiii exposure with no moon is not equal to an hour of Oiii exposure with the moon up. An hour of exposure with the moon 45 degrees away is not equal to an hour of exposure with the moon 180 away. And so on and so forth

Edited by gsaramet, 19 April 2025 - 02:15 PM.

[JWolf_21]

It can easily get more complicated Consider a 5x5 = 25 panel mosaic with 20% overlap and a 10 hour integration per panel. Let it be a color camera.  What is the integration time?

 

Now extrapolate to SHO image with different times per channel AND RGB for star colors

 

And this is just math tricks. The REAL McCoy is this: 

 

One hour of a channel is NOT equal to one hour of the same channel. An hour of exposure at 15 degrees altitude is not the same as an hour of exposure at zenith. An hour of exposure in Bortle 8 is not equal to an hour of exposure in Bortle 3. An hour of Oiii exposure with no moon is not equal to an hour of Oiii exposure with the moon up. An hour of exposure with the moon 45 degrees away is not equal to an hour of exposure with the moon 180 away. And so on and so forth

Exactly, so even giving exposure time per filter is inaccurate enough... no need to give another parameter (the total integration time) that only says how much time you spent watching netflix while your equipment was doing its job From now on, the total integration time should be renamed to "total waiting time"  

[gsaramet]

 From now on, the total integration time should be renamed to "total waiting time" 

Beg your pardon, the total waiting time is the time you wait for the images to be stacked

[Alex McConahay]

From JWolf 21, Post #1 >>>>>>>I'm used to hear to some of my astrophotography friends, and many internet posts, say "the exposure time of this image was 9 hours.... 3 hours of Ha, 3 hours of OIII and 3 hours of NII" or whatever filters they are using, RGB, whatever...

 

From my point of view, the exposure time of the full picture should be 3 hours, not 9,"

 

Words and language mean what we all have come to agree what they mean. 

 

The amateur astroimaging world has come to use the term "exposure time" to mean the "total time the camera was gathering light on the target...." regardless of which filter was being used, or the transmission of the filter, or sensor quantum efficiency, or sensor type (LRGB, OSC, or Ektachrome), or how many of the exposures were thrown out for poor tracking/guiding, bad seeing, bad whatever...... before stacking.   It is less a measure of photons captured in whatever color as it is a measure of how long the shutter was open, how long the imager spent gathering data. 

 

Your question is interesting. And you are right in your analysis.

 

But there is no need to take a perfectly good phrase ("total exposure") and try to redefine it for some other purpose. Coin a new phrase if you like. But "total exposure" already has a meaning. 

 

Alex

[UP4014Fan]

Exactly, so even giving exposure time per filter is inaccurate enough... no need to give another parameter (the total integration time) that only says how much time you spent watching netflix while your equipment was doing its job From now on, the total integration time should be renamed to "total waiting time"

I don’t necessarily agree with this. Having some inkling of conditions and exposure helps others reasonably target where to at least start. Going back to my terrestrial example, after years of experience and knowing what worked for others, I knew that 30 seconds for the church and 5 seconds for the foreground would work (and don’t think I didn’t bracket these).

For myself as a newbie in digital AP, having read and seen that somewhere around 30 seconds to a minute per subframe with a Canon DSLR with a minimum of three hours integration time gave me a baseline with which to start to learn (interestingly, the NINA exposure plug-in says that my camera’s ideal exposure time in my skies is about 31 seconds).

No doubt I’m doing other things while monitoring my set up. While I remote into my laptop that is driving the imaging session, I’m watching movies or TV, doing the Washington “Post” crossword puzzle, or whatnot on the other two monitors. All the while being grateful that I don’t have to manually guide the scope (I’m in awe of the astronomers who spent hours at guide scopes in the old days did that!). That doesn’t mean the time I’m spending is dead time.

Maybe some sort of way of measuring overall ADU or photon count is the ticket? And thinking of this, what measurement has value when doing this using film, where plates were exposed over multiple nights for hours on end?

Edited by UP4014Fan, 19 April 2025 - 03:46 PM.

[Zambiadarkskies]

It can easily get more complicated Consider a 5x5 = 25 panel mosaic with 20% overlap and a 10 hour integration per panel. Let it be a color camera.  What is the integration time?

 

Now extrapolate to SHO image with different times per channel AND RGB for star colors

 

And this is just math tricks. The REAL McCoy is this: 

 

One hour of a channel is NOT equal to one hour of the same channel. An hour of exposure at 15 degrees altitude is not the same as an hour of exposure at zenith. An hour of exposure in Bortle 8 is not equal to an hour of exposure in Bortle 3. An hour of Oiii exposure with no moon is not equal to an hour of Oiii exposure with the moon up. An hour of exposure with the moon 45 degrees away is not equal to an hour of exposure with the moon 180 away. And so on and so forth

This.  Exposure time is the time (in total) that you expose your sensor that then ends up in the final stack.  To try and break it down to "reality" is pointless.  An overcomplication that an already complex endeavor just doesn't need.  Where to draw the line? And what about focal ratio?  

[JWolf_21]

This.  Exposure time is the time (in total) that you expose your sensor that then ends up in the final stack.  To try and break it down to "reality" is pointless.  An overcomplication that an already complex endeavor just doesn't need.  Where to draw the line? And what about focal ratio?  

I actually find more sense in knowing the exposure time per filter than the total exposure time, because the first is informative while the second says nothing.

 

For instance, somebody says he took a picture on HSO palette of 9 hours of total exposure time, and I want to get a similar result... should I use 8 hours of H, 30 min S and 30 min O? or 3 hours each? or any other combination? Knowing that he used his camera for 9 hours tells nothing about the final picture. I would not know how much time I should put on each channel and I'd do a mess.

 

The total exposure time or total integration time, when stacking different wavelengths, is just a measure of how long took the full procedure, but says nothing technically interesting about the picture that I could use to replicate or improve the result.

 

I don't see any complication, if you report that your image had X hours per channel. If there's a complication, if any, it would be to sum those hours to give an additional value that is meaningless (except for the usage count of your sensor).

Edited by JWolf_21, 19 April 2025 - 03:44 PM.

[JWolf_21]

I don’t necessarily agree with this. Having some inkling of conditions and exposure helps others reasonably target where to at least start. Going back to my terrestrial example, after years of experience and knowing what worked for others, I knew that 30 seconds for the church and 5 seconds for the foreground would work (and don’t think I didn’t bracket these).

For myself as a newbie in digital AP, having read and seen that somewhere around 30 seconds to a minute per subframe with a Canon DSLR with a minimum of three hours integration time gave me a baseline with which to start to learn (interestingly, the NINA exposure plug-in says that my camera’s ideal exposure time in my skies is about 31 seconds).

No doubt I’m doing other things while monitoring my set up. While I remote into my laptop that is driving the imaging session, I’m watching movies or TV, doing the Washington “Post” crossword puzzle, or whatnot on the other two monitors. All the while being grateful that I don’t have to manually guide the scope (I’m in awe of the astronomers how spent hours at guide scopes in the old days did that!). That doesn’t mean the time I’m spending is dead time.

Maybe some sort of way of measuring overall ADU or photon count is the ticket? And thinking of this, what measurement has value when doing this using film, where plates were exposed over multiple nights for hours on end?

I see your point, but you're talking about a DSLR, which is a different story. If you tell me you used 3 hours for an object on your color DSLR, I can use 3 hours on mine too and we might get the same result. But here it's a single RGB shot or sum of them.

 

However, when you are collecting separate channels, and with different exposure times each, the story is a different one. Because if you tell me you needed 3 hours for an object with a mono camera and R, G and B filters... how would I know how much time to put on each filter? so your reference of 3 hours would not be as useful, since I would not know how to split that time on each channel.

[JWolf_21]

But there is no need to take a perfectly good phrase ("total exposure") and try to redefine it for some other purpose. Coin a new phrase if you like. But "total exposure" already has a meaning. 

 

Alex

Sure, I just opened this topic because I was unsure whether using the total stacking time was relevant for a picture taken by channels or not. Precisely, I think I understand the meaning of total exposure and my friends don't, that's why they only use that parameter as a synonym of how good their pictures are, since "the more exposure time the better", while I think the total exposure is an oversimplification and very little informative about a stacked image.

[Zambiadarkskies]

I actually find more sense in knowing the exposure time per filter than the total exposure time, because the first is informative while the second says nothing.

For instance, somebody says he took a picture on HSO palette of 9 hours of total exposure time, and I want to get a similar result... should I use 8 hours of H, 30 min S and 30 min O? or 3 hours each? or any other combination? Knowing that he used his camera for 9 hours tells nothing about the final picture. I would not know how much time I should put on each channel and I'd do a mess.

The total exposure time or total integration time, when stacking different wavelengths, is just a measure of how long took the full procedure, but says nothing technically interesting about the picture that I could use to replicate or improve the result.

I don't see any complication, if you report that your image had X hours per channel. If there's a complication, if any, it would be to sum those hours to give an additional value that is meaningless (except for the usage count of your sensor).

So what about focal ratio as I mentioned? At the end of the day for me it is end results that matter. The rest is just details...

Edit: half the images shared on here don't even mention sky quality, location and gear. Let alone integration time. Let alone filter handicaps and other scales. I admire your intentions and effort. I prefer to shout at clouds. Literally.

Edited by Zambiadarkskies, 19 April 2025 - 03:56 PM.

[JWolf_21]

So what about focal ratio as I mentioned? At the end of the day for me it is end results that matter. The rest is just details...

Of course there are more parameters to consider when analyzing images, but I was just asking about integration time on a simple example, assuming the same equipment, so all other parameters remain invariant.

[UP4014Fan]

I see your point, but you're talking about a DSLR, which is a different story. If you tell me you used 3 hours for an object on your color DSLR, I can use 3 hours on mine too and we might get the same result. But here it's a single RGB shot or sum of them.

However, when you are collecting separate channels, and with different exposure times each, the story is a different one. Because if you tell me you needed 3 hours for an object with a mono camera and R, G and B filters... how would I know how much time to put on each filter? so your reference of 3 hours would not be as useful, since I would not know how to split that time on each channel.

Which is why I’m suggesting a common denominator, to which someone has already alluded. Total photons? Average photons? Average or median ADU from the master? At the end of the day, we wind up with an RGB (or maybe gray scale) image, so I dunno…

[jamku]

Say I take a photo with a dslr, set shutter time to 4 seconds, and the photo is of a blue surface. It's going to be a four second exposure even though only 1/4 of the pixels are doing much and even if the source is a narrow spectrum. I see your idea that exposure time would be as part of the full spectrum/all three bands - but what is a full spectrum to begin with, not every photo necessarily contains all three r, g, b at all, all this is needless convolution of the fact that duration of the exposure was four seconds.

 

So, JWolf_21,  I'm with your buddies on this one

[Zambiadarkskies]

Of course there are more parameters to consider when analyzing images, but I was just asking about integration time on a simple example, assuming the same equipment, so all other parameters remain invariant.

Focal ratio is intrinsic to integration time. If you just ignore it on the basis that it is another parameter that remains invariant the I just don't understand the whole argument. What about target brightness? Is that also just another parameter that should remain invariant? Pixel size too? Other sensor parameters such as full wheel depth? This is my point. There are so many variables that it sort of becomes futile IMHO.

[JWolf_21]

Which is why I’m suggesting a common denominator, to which someone has already alluded. Total photons? Average photons? Average or median ADU from the master? At the end of the day, we wind up with an RGB (or maybe gray scale) image, so I dunno…

Simply the exposure time per filter would be enough (provided that I know which aperture, FL and camera you're using, so I can adapt those times to mine equipment).

[JWolf_21]

Say I take a photo with a dslr, set shutter time to 4 seconds, and the photo is of a blue surface. It's going to be a four second exposure even though only 1/4 of the pixels are doing much and even if the source is a narrow spectrum. I see your idea that exposure time would be as part of the full spectrum/all three bands - but what is a full spectrum to begin with, not every photo necessarily contains all three r, g, b at all, all this is needless convolution of the fact that duration of the exposure was four seconds.

 

So, JWolf_21,  I'm with your buddies on this one

It's 1/4 pixels, ok, you lose resolution... but the exposure time was 4 seconds in each channel (you just only captured blue photons on a blue channel on 1/4 of the sensor). Right? You don't say you exposed 12 s (4 per channel...)

Edited by JWolf_21, 19 April 2025 - 04:05 PM.