This is the clearest explanation I’ve heard of this issue, and you’ve helped me correct some of my own flawed understanding. I look forward to trying out a lower ISO to improve my photos. Thanks.
Really awesome video! Thanks for taking the time and effort to explain the ISO aspect and especially the dual-gain type since I have seen that graph on the photons website but did not clue in as to what's going on and some of the people who comments on the forum gave poor explanations.
Thank you so much for explaining how these things work, I knew the terms but was unsure how they related to each other.. now I know.. thanks very much.
Thanks, RoRo for an informative video on gain. I would love to see a video on offset for dedicated astronomy cameras. I recently reviewed part 1 of this and agree with your suggested exposure times.
Thanks for your video, RoRo, and the great question from Ron Stewart. I've also got a Sony, the very similar but cheaper a6000. I use it on the back of a GSO 8" Ritchey Chretien on an iOptron AzPro mount. I've got a very small sky window so I can't rely on having enough guide stars available to polar align a proper EQ mount. I am limilted to 30 - 40 sec subframe exposures to keep the radial star smearing controlled (due to the 15 deg per hour rotation of a tracked object when using an Alt-Az mount). So far I have been using high ISOs of 3200 and 6400 on the Horesehead nebula from my Bortle 5 backyard in suburban Melbourne and so was surprised that the optimum ISO for my camera is 400. I had similar problems to Ron with it being barely perceptable against the background noise. If I try using ISO 400 could I get something to process if I were to use lots and lots of subframes? (Say 500 instead of the usual 30 to 40, given that I will be getting one sixteenth of the previous brightness)
Hi RoRo. Great video. You're video is the first I've encountered that explains so clearly the effects of different gain settings. I have a ZWO ASI2600MC Pro, and from the Read Noise and Dynamic Range graphs, it is quite obvious that it is, as you called it, a dual gain sensor, because of the huge drop in RN and uptick in DR. Based on that, it appears that a gain of 100 is optimal for this camera. (I hope I got that right?). Having said that, based on the graphs that ZWO has posted on their website for this camera, at gain 100 the Full Well depth drops, not greatly but certainly noticeably. The graph isn't clear on just exactly where FW sits at with gain set at 100. I am a bit concerned how DR will be affected. There is another graph called GAIN-e/ADU but I do not understand its significance and possible effects at the 100 gain setting. Are my concerns regarding FW and GAIN-e/ADU valid? Btw, I did see your previous video on the ideal exposure settings. Very helpful!!! Your explanations are more in depth and easy to understand. Please keep them coming. Thank you!
I'm finding these videos really helpful, and I can see that I've been using way too high a gain setting on my ZWO ASI 294. Many thanks for this. Not sure if anyone else has mentioned this, but can I offer a quick correction that a decibel is 1/10th of a bel (at 4:05 you say it's the other way around)
Awesome video!! thank you for clarifying lots of misconceptions.. i got my a7siii astro moded for stills of milkyway and some deep space stuff.. what would be your recommended settings i should be using if i got a star tracker.. i am mainly concerned with the iso settings range.. i feel 1600 would blow out highlights v fast but keeping iso lower than 1600will add lots of grain .. could you kindly help here
Nice vid, useful information. My analog gain values in the ASCOM driver go from 0-50, whereas it goes from 0-2000 in the digital gain setting in the main software. In APT, if I leave gain blank, it links to the ASCOM anaog gain, where a gain of 10 translates to 290 out of 2000 in the software (thanks APT for the running commentary on the bottom of the programme). One query I have been testing and wondering: for dual, tri or quad band filters with an OSC sensor, is dynamic range important since the bandwidths limit the spectral range by definition. Are higher gain (compensating for filter opacity anyway) also useful for brightness (without oversampling or saturating stars) where dynamic range is already reduced tot he bandpass of the filter? Can I focus on well depth and less on dynamic range with narrow or multibandpass filters?
I think you should put out another video and explain how stacking effects dynamic range. The bit depth/full well only needs to be big enough so that how many photos you are willing to stack will push its final bit depth and dynamic range to something acceptable for what you are shooting.
Hi there can you tell me what gain and offset would be good for on my ZWO 1600 mm pro when I’m using LRGB and when I’m using my 7nm Ha Oiii and Sii filters? My skies are a bortle 6… I also watched your other video on short exposures but how about for a 7 nm filter? Thank you very much you have awesome tutorials keep up the awesome images and awesome videos🔭🔭
I think using ISO 100 is not a good option for most of the situations, ( when you do not take very long exposures) because you will not have enough signal to work with after stacking the image. I have tested this with 1 min exposures
Thanks a lot for this specialist talk. I am a bit surprised by the statement though. I was under the impression that the important question is how the squared read noise varies comparatively to full well capacity when gain increases. Basically, I would have assumed that if full well capacity decreases faster than read noise squared, then the final stack (at equal total exposure time) will either have more noise or more clipped highlights.
The idea is that combining this knowledge with the ideas in my last video around exposure length we can use much higher gain and much lower exposures than most people think. Of course, given excellent equipment and conditions, we will still want to expose for long times with lower gain, but not everyone has those perfect conditions. If you're already limited to say 45 second images due to gear, then increasing your gain will probably help you more than hurt since your exposures are so short. It's all a balancing act. The cheat sheets are for those who want to find a good starting point, and the rest of the video is for those who want to find out more about why so they can make more informed decisions about their own setups :)
@@AstroWithRoRo perhaps the applicability depends on the camera. Also, specific examples with real numbers could help understand when higher gain could be detrimental. For instance, if we take Sharpcap sub length formula C*R^2/P with C=10 (5% extra noise target) and P=2.6 (Bortle 5, f/6) and use the ASI071MC pro. At native gain, read noise R=3.3e and sub length is 10*3.3^2/2.6=42 seconds. At unity gain, R=2.6e and sub length becomes 10*2.6^2/2.6=26s which is 1.6 times smaller than the previous 42 seconds. Now, if we look at full well, at native gain, FW=48.6K but at unity gain, FW=16K which is 3 times smaller. With these numbers, all the pixels that were more than 53% full in the 42 seconds exposure will be clipped in the 26s exposure. To avoid that clipping at unit gain, you would need to reduce the sub length to 14s or so, which would increase the noise in the resulting stack. In this scenario, increasing gain seems counterproductive from the point of view of noise (and dynamic range). On the other hand, with the ASI533MC pro which has better ADC and amp circuits, it becomes a no-brainer to use unity gain instead of native gain because read noise improves a lot more (from 3.7e to 1.5e) , leading to sub lengths of 53 and 9 seconds respectively, while FW has similar reduction (from 50k to 17K). In this case, the pixels that were 99% full on the 53s sub at native gain are only 50% full on the 9s sub at unity gain. Conversely, this enables an 18s sub at unity gain that will not have more clipped pixels than the 53s sub at native gain. This 18s sub length will lead to less extra noise in the resulting stack (C=20 instead of C=10, which would be 3% instead of 5% extra noise).
quick googling shows that my 6D is 25%-50% QE LOL i guess there is a reason there are dedicated astro cams for deep sky! still 6D great for landscape, tracked astro. Is there a way to calculate QE? I'd be curious to be able to calculate this for various mirrorless cameras and DSLRs for use in SERIOUS landscape astrophotography.
Awesome. So right now I, a beginner, am trying to capture the Horsehead Nebula. The Ha around the head is barely brighter than the noise, and the head is extremely noisy. My Sony a6100 has a step in the curve that screams that I should use ISO 400. Until I get an external intervalometer, I am stuck with 30-second subs. Bortle 6, Astronomik CLS filter. So, is there any point in experimenting with higher ISO?
Since you're limited to 30s exposures anyway due to the internal camera settings. I'd absolutely try higher ISOs, at 30s you should be able top push the ISO a bit before you find any clipping or DR issues. Stacking extra images can also give you back more dynamic range, so you'll have many subs which will help ease the lack of DR reduction too.
The noise is caused by underexposure or by raising the brightness in post. Low light causes high noise. Regardless of the ISO, regardless of the exposure, pictures with less light will have more noise than pictures with more light. Therefore, if you want to reduce noise, add light!
The amount of light does *NOT* change the read noise. As shown in the video, the read noise changes with ISO or Gain. Yes, it is possible to increase the *SNR* by increasing the amount of light (signal). But no you do not decrease the read noise, by increasing the light. Did you the watch video? Did you understand what was said?
So if I am understanding this, changing the iso doesn't change the sensitivity of the sensor at all? If I take a picture and increase iso it's the same as putting that low iso pic on a computer and increasing the "exposure" in post.
On the 1600MM 100 gain is the point where the read noise starts to flatten out and dynamic range really starts to drop. The full well depth is also still decent. You can also use 50 gain as a more conservative number for more FWD.
@@AstroWithRoRo Okay thanks, More darks to capture then while its cloudy and raining. I captured dark libraries at unity gain (works out at 138 for my sensor on my 1600mm).
You seem to be confused about Bels and decibels dB's. There are 10 dB's in a Bel, the clue is in the lowercase d which denotes deci is 1/10th. Have a look at en.wikipedia.org/wiki/Decibel I've found I can use my Canon Ra up to ISO6400 easily without noise being a problem.
May I ask why you said when there was only one photon came in it is then registered between 0 and 5? I thought it depends on the brightness of the photon. The darker the photon the lower the number? Thank you.
AFIK every photon is the same brightness, they only difference is their wavelength which determines the colour. The brightness in an image is determined by the number of photons hitting a pixel. The more photons, the brighter, less photons = more faint. There's also a slight randomness added to each pixel readout though (this is the read noise) which is any it could be a bit more or less than the actual photon/electron reading.
@@AstroWithRoRo I see. So should it be the wavelength of a photon determines where it sits between 1 and 65535 on a 16 bit? Just wanna understand the logic. Thanks.
Exactly! For OSC cameras, just aim for an overall middle ADU count. For mono cameras you will need to adjust the brightness for each filter to make sure it's exposed correctly for the different wavelengths.
I'm still confused as hell. You are saying what I've heard other people say - that is, the same person will say higher ISO results in more noise (and this is how I understand it, and it's what I see in my pictures), then that same person apparently contradicts themself by saying that increasing ISO actually results in less noise (Huh? Which is it?) Somewhere in this discussion there is some crucial point that I'm missing. What is that thing??? I've watched your video over and over and I can't discover what everybody else seems to be getting.
Not true. Read noise is not a significant factor for most application And increasing gain above unity (1 ev / photon) is of no value at all. It is like magnification beyond resolution and can saturate your sensor. Increasing gain generally reduces dynamic range.
This is the clearest explanation I’ve heard of this issue, and you’ve helped me correct some of my own flawed understanding. I look forward to trying out a lower ISO to improve my photos. Thanks.
Really awesome video! Thanks for taking the time and effort to explain the ISO aspect and especially the dual-gain type since I have seen that graph on the photons website but did not clue in as to what's going on and some of the people who comments on the forum gave poor explanations.
This guy knows what he is talking about, thanks RoRo.
Absolute legend. Such a clear & concise explanation. 🙂
Great analysis. Really well done
Terrific explanation of these concepts! Thanks so much for a fantastic video. This has been a bit of a mystery to me - until now!
Glad you enjoyed it!
Another excellent, clear explanation!
Great explanation, Thanks!
Thank you so much for explaining how these things work, I knew the terms but was unsure how they related to each other.. now I know.. thanks very much.
Thanks, RoRo for an informative video on gain. I would love to see a video on offset for dedicated astronomy cameras. I recently reviewed part 1 of this and agree with your suggested exposure times.
Thanks for your video, RoRo, and the great question from Ron Stewart. I've also got a Sony, the very similar but cheaper a6000. I use it on the back of a GSO 8" Ritchey Chretien on an iOptron AzPro mount. I've got a very small sky window so I can't rely on having enough guide stars available to polar align a proper EQ mount. I am limilted to 30 - 40 sec subframe exposures to keep the radial star smearing controlled (due to the 15 deg per hour rotation of a tracked object when using an Alt-Az mount).
So far I have been using high ISOs of 3200 and 6400 on the Horesehead nebula from my Bortle 5 backyard in suburban Melbourne and so was surprised that the optimum ISO for my camera is 400. I had similar problems to Ron with it being barely perceptable against the background noise.
If I try using ISO 400 could I get something to process if I were to use lots and lots of subframes? (Say 500 instead of the usual 30 to 40, given that I will be getting one sixteenth of the previous brightness)
Very informative 👍. Thanks for sharing..
Great video! Happy I came across your channel. Glad to know I was reading the charts right 😂
Thanks for stopping by, glad the video helped! :)
great channel. very simplified from complicated things :)
Glad you liked it!
Excellent explanations. I’m a subscriber now.
Thanks and welcome!
Super helpful. Thanks
Man your videos are awesome!
Thank you!
Very good video :)
yo, good stuff. im struggling with this now, and this should help me out.
Thanks! I hope it helps :)
Surprised my 60D made it in the list. Good stuff!
Me too! However i always hear that ISO 1600 is the best for it. I was very surprised that it's 640. ?
Hi RoRo. Great video. You're video is the first I've encountered that explains so clearly the effects of different gain settings. I have a ZWO ASI2600MC Pro, and from the Read Noise and Dynamic Range graphs, it is quite obvious that it is, as you called it, a dual gain sensor, because of the huge drop in RN and uptick in DR. Based on that, it appears that a gain of 100 is optimal for this camera. (I hope I got that right?). Having said that, based on the graphs that ZWO has posted on their website for this camera, at gain 100 the Full Well depth drops, not greatly but certainly noticeably. The graph isn't clear on just exactly where FW sits at with gain set at 100. I am a bit concerned how DR will be affected. There is another graph called GAIN-e/ADU but I do not understand its significance and possible effects at the 100 gain setting. Are my concerns regarding FW and GAIN-e/ADU valid? Btw, I did see your previous video on the ideal exposure settings. Very helpful!!! Your explanations are more in depth and easy to understand. Please keep them coming. Thank you!
I'm finding these videos really helpful, and I can see that I've been using way too high a gain setting on my ZWO ASI 294. Many thanks for this. Not sure if anyone else has mentioned this, but can I offer a quick correction that a decibel is 1/10th of a bel (at 4:05 you say it's the other way around)
Awesome video!! thank you for clarifying lots of misconceptions.. i got my a7siii astro moded for stills of milkyway and some deep space stuff.. what would be your recommended settings i should be using if i got a star tracker.. i am mainly concerned with the iso settings range.. i feel 1600 would blow out highlights v fast but keeping iso lower than 1600will add lots of grain .. could you kindly help here
ZWO uses Unity gain as the balanced point of gain and Dynamic range, eg 294mc unity gain is at 120 Thanks for the video
You got it!
Nice vid, useful information. My analog gain values in the ASCOM driver go from 0-50, whereas it goes from 0-2000 in the digital gain setting in the main software. In APT, if I leave gain blank, it links to the ASCOM anaog gain, where a gain of 10 translates to 290 out of 2000 in the software (thanks APT for the running commentary on the bottom of the programme).
One query I have been testing and wondering: for dual, tri or quad band filters with an OSC sensor, is dynamic range important since the bandwidths limit the spectral range by definition. Are higher gain (compensating for filter opacity anyway) also useful for brightness (without oversampling or saturating stars) where dynamic range is already reduced tot he bandpass of the filter? Can I focus on well depth and less on dynamic range with narrow or multibandpass filters?
I think you should put out another video and explain how stacking effects dynamic range. The bit depth/full well only needs to be big enough so that how many photos you are willing to stack will push its final bit depth and dynamic range to something acceptable for what you are shooting.
Bit depth in the stacking program will be at least 32 bits.
@@ThinkingBiblically A file can appear as a higher bit depth, but it can also just be interpolation. The true dynamic range can be less than that.
I have an A7 III... Winning!
Hi there can you tell me what gain and offset would be good for on my ZWO 1600 mm pro when I’m using LRGB and when I’m using my 7nm Ha Oiii and Sii filters? My skies are a bortle 6… I also watched your other video on short exposures but how about for a 7 nm filter? Thank you very much you have awesome tutorials keep up the awesome images and awesome videos🔭🔭
How do you not have 200,000 subs?
I think using ISO 100 is not a good option for most of the situations, ( when you do not take very long exposures) because you will not have enough signal to work with after stacking the image. I have tested this with 1 min exposures
Thanks a lot for this specialist talk. I am a bit surprised by the statement though. I was under the impression that the important question is how the squared read noise varies comparatively to full well capacity when gain increases. Basically, I would have assumed that if full well capacity decreases faster than read noise squared, then the final stack (at equal total exposure time) will either have more noise or more clipped highlights.
As I understand it, the rules of thumb Rowan gives find the spot where the conditions you state tend to exist.
The idea is that combining this knowledge with the ideas in my last video around exposure length we can use much higher gain and much lower exposures than most people think. Of course, given excellent equipment and conditions, we will still want to expose for long times with lower gain, but not everyone has those perfect conditions. If you're already limited to say 45 second images due to gear, then increasing your gain will probably help you more than hurt since your exposures are so short. It's all a balancing act. The cheat sheets are for those who want to find a good starting point, and the rest of the video is for those who want to find out more about why so they can make more informed decisions about their own setups :)
@@AstroWithRoRo perhaps the applicability depends on the camera. Also, specific examples with real numbers could help understand when higher gain could be detrimental. For instance, if we take Sharpcap sub length formula C*R^2/P with C=10 (5% extra noise target) and P=2.6 (Bortle 5, f/6) and use the ASI071MC pro. At native gain, read noise R=3.3e and sub length is 10*3.3^2/2.6=42 seconds. At unity gain, R=2.6e and sub length becomes 10*2.6^2/2.6=26s which is 1.6 times smaller than the previous 42 seconds. Now, if we look at full well, at native gain, FW=48.6K but at unity gain, FW=16K which is 3 times smaller. With these numbers, all the pixels that were more than 53% full in the 42 seconds exposure will be clipped in the 26s exposure. To avoid that clipping at unit gain, you would need to reduce the sub length to 14s or so, which would increase the noise in the resulting stack. In this scenario, increasing gain seems counterproductive from the point of view of noise (and dynamic range). On the other hand, with the ASI533MC pro which has better ADC and amp circuits, it becomes a no-brainer to use unity gain instead of native gain because read noise improves a lot more (from 3.7e to 1.5e) , leading to sub lengths of 53 and 9 seconds respectively, while FW has similar reduction (from 50k to 17K). In this case, the pixels that were 99% full on the 53s sub at native gain are only 50% full on the 9s sub at unity gain. Conversely, this enables an 18s sub at unity gain that will not have more clipped pixels than the 53s sub at native gain. This 18s sub length will lead to less extra noise in the resulting stack (C=20 instead of C=10, which would be 3% instead of 5% extra noise).
Great call out. I may have to do a more detailed video around over-exposure to cover all this and really clarify it. :)
another question... how would we know that we clipped our highlights if were going for milkyway and deep space imaging?
quick googling shows that my 6D is 25%-50% QE LOL i guess there is a reason there are dedicated astro cams for deep sky! still 6D great for landscape, tracked astro. Is there a way to calculate QE? I'd be curious to be able to calculate this for various mirrorless cameras and DSLRs for use in SERIOUS landscape astrophotography.
Awesome. So right now I, a beginner, am trying to capture the Horsehead Nebula. The Ha around the head is barely brighter than the noise, and the head is extremely noisy. My Sony a6100 has a step in the curve that screams that I should use ISO 400. Until I get an external intervalometer, I am stuck with 30-second subs. Bortle 6, Astronomik CLS filter. So, is there any point in experimenting with higher ISO?
Since you're limited to 30s exposures anyway due to the internal camera settings. I'd absolutely try higher ISOs, at 30s you should be able top push the ISO a bit before you find any clipping or DR issues. Stacking extra images can also give you back more dynamic range, so you'll have many subs which will help ease the lack of DR reduction too.
@@AstroWithRoRo Thank you. Too bad we didn't have this exchange 30 hours ago. The clouds have rolled in until the full moon.
The noise is caused by underexposure or by raising the brightness in post. Low light causes high noise. Regardless of the ISO, regardless of the exposure, pictures with less light will have more noise than pictures with more light. Therefore, if you want to reduce noise, add light!
The amount of light does *NOT* change the read noise. As shown in the video, the read noise changes with ISO or Gain. Yes, it is possible to increase the *SNR* by increasing the amount of light (signal). But no you do not decrease the read noise, by increasing the light. Did you the watch video? Did you understand what was said?
So if I am understanding this, changing the iso doesn't change the sensitivity of the sensor at all? If I take a picture and increase iso it's the same as putting that low iso pic on a computer and increasing the "exposure" in post.
If your camera is ISO invariant then yes.
0-65,335
0-65,535, but yeah, the off-by-one is the significant bit, so to speak.
Maybe I'm missing something, but I can't work out how you arrived at gain 100 for the 1600mm.
On the 1600MM 100 gain is the point where the read noise starts to flatten out and dynamic range really starts to drop. The full well depth is also still decent. You can also use 50 gain as a more conservative number for more FWD.
@@AstroWithRoRo Okay thanks, More darks to capture then while its cloudy and raining. I captured dark libraries at unity gain (works out at 138 for my sensor on my 1600mm).
You seem to be confused about Bels and decibels dB's. There are 10 dB's in a Bel, the clue is in the lowercase d which denotes deci is 1/10th. Have a look at en.wikipedia.org/wiki/Decibel
I've found I can use my Canon Ra up to ISO6400 easily without noise being a problem.
May I ask why you said when there was only one photon came in it is then registered between 0 and 5? I thought it depends on the brightness of the photon. The darker the photon the lower the number? Thank you.
AFIK every photon is the same brightness, they only difference is their wavelength which determines the colour. The brightness in an image is determined by the number of photons hitting a pixel. The more photons, the brighter, less photons = more faint. There's also a slight randomness added to each pixel readout though (this is the read noise) which is any it could be a bit more or less than the actual photon/electron reading.
@@AstroWithRoRo I see. So should it be the wavelength of a photon determines where it sits between 1 and 65535 on a 16 bit? Just wanna understand the logic. Thanks.
Exactly! For OSC cameras, just aim for an overall middle ADU count. For mono cameras you will need to adjust the brightness for each filter to make sure it's exposed correctly for the different wavelengths.
04:11 I thought an increase of +3db doubles the amplitude?
An increase of 3db doubles the power ratio, but 6db doubles the amplitude.
@@AstroWithRoRo Thanks for clarifying :-)
I'm still confused as hell. You are saying what I've heard other people say - that is, the same person will say higher ISO results in more noise (and this is how I understand it, and it's what I see in my pictures), then that same person apparently contradicts themself by saying that increasing ISO actually results in less noise (Huh? Which is it?)
Somewhere in this discussion there is some crucial point that I'm missing. What is that thing??? I've watched your video over and over and I can't discover what everybody else seems to be getting.
Not true. Read noise is not a significant factor for most application And increasing gain above unity (1 ev / photon) is of no value at all. It is like magnification beyond resolution and can saturate your sensor. Increasing gain generally reduces dynamic range.
Let's amend this to "High ISO doesn't _necessarily_ equal higher noise". Otherwise it's clickbait and that puts people like me off.