This is one of the best explanations I’ve seen related to noise and optimal exposure length! It was so easy to follow that I can almost pretend my brain is 15 years younger. Long, dry, boring purely technical explanations of these complex concepts make me feel like I need to lie down and take a nap before my 4:00 PM early-bird special buffet dinner. This was not one of those, and I appreciate it!
Great stuff. Very different from many astronomy channels I follow. Nice to see some deeper explanations than just seeing unboxing and superficial review videos. Looking forward to see the next episodes.
Thank you David - once I manage to build a time machine, I'll make sure to go back 32 years into the past and replace your instructor for the introduction to the topic ;) That's awesome though, did you end up working in the field you studied?
@@CuivTheLazyGeek I became a medical physicist, but instead of working in diagnostic imaging I work in therapeutic radiological physics, radiation oncology. I still use much of my knowledge of digital imaging in my work, but most of my passion for imaging is in my hobby. Thanks for producing these videos. You have a wonderful way of explaining things.
NINA made me laugh tonight. I was reading through the release notes and one entry stood out. Various grammar and SPELING errors corrected :) With respect to my question on Celestron mount parking, it is looking like a bug in the Celestron driver as APT has the same problem.
Hahaha that is pretty cool! Almost wondering whether it has been inserted intentionally :) I should ask around. Bummer on the Celestron thing.... I hope it gets fixed - I used to have an account on their test/dev forum, and the devs were very reactive, so it may be worth asking.
Great explanation! At some point I thought you could have made this video shorter and say the same, but I couldn't say it was boring. Thanks for making a quite complicated item clear with simple words.
So I have an off the wall question for you, but you might be able to help my understanding here. I haven't made it through this series of videos, but my question is a bit different. I understand the basics of "Shot" noise and "Read" noise (Upstream which can be amplified by dialing up amplification with ISO, and downstream - inherent to the camera). My question is this (sort of has to do with stacking - sort of). A well known photographer will "stack/combine" three 14-bit images in Photoshop (i.e. a portrait containing environment). He will photograph the scene using a strobe. He brackets his the three shots (not going to use the word exposure here) by only varying ISO. So flash power, aperture, and shutter speed remain constant. His goal is a 32-bit file in the largest color space to manipulate and print. His work is excellent, but I have a question regarding the science of his bracketing technique -- which he either can't or won't explain. In an ISO invariant sensor, is there any advantage of bracketing ISO two stops up and down or center vs taking the same image, copying it into three images, and combining those and then pushing the processing of highlights and shadows? I realize, that by copying the image twice, you eliminate the variance of environment "shot noise" differences, but do you think you can actually see the difference? It seems, by bracketing ISO, you're just introducing more noise then needed - certainly at the higher ISO. I do realize that some cameras (i.e. Nikon D850 has two processing engines that gives ISO 400 more DR than ISO 380). That aside, can you see any advantage to doing a bracket of ISO when keeping the other variables of SS/Aperture/flash power unchanged?
Well, ISO is just an amplifier. Higher ISO isn't intrinsically noisier than lower ISO in terms of SNR. In fact, higher ISOs typically have LOWER read noise (very important for astrophotography), but sacrifice dynamic range to some extent. Higher ISOs appear to have more noise, simply because normally higher ISO is associated with something like faster shutter speed (to freeze the moment for instance), but in that case everything, including the shot noise, gets amplified by the high ISO to achieve the same brightness level as a longer exposure with lower ISO. With a shorter exposure (regardless of ISO) you will achieve less signal, less shot noise (sqrt(signal)) and worse Signal to Shot Noise Ratio than a longer exposure. If you used higher ISO to achieve higher overall brightness you'll have LESS read noise, but also less dynamic range than an exposure with lower ISO. With bracketing you have: - an underexposed low ISO image with high dynamic range and high read noise (read noise is usually inconsequential in well lit conditions, but very important in astrophotography) - a well exposed normal ISO image with good DR and "normal" read noise (sweet spot is usually around 800 for a lot of cameras in terms of read noise). Maybe the normally exposed image could have clipped highlights somewhere. - an overexposed high ISO image with low dynamic range, potentially clipped highlights, low read noise. Because all share the same parameters besides the ISO (which is just an amplifier, nothing more nothing less), they have sensibly the same SNR (low ISO slightly worse because of high read noise). But each covers a different dynamic range (say 14 stops for low iso image, 13 stops for medium iso image, 12 stops for high iso image), centered differently. Depending on how the bracketed images are combined (as assume some kind of HDR replacement + averaging), you will have better SNR by increasing signal more than shot noise, because square root function (although the LOW ISO image does introduce slightly more read noise), AND an HDR picture that covers a lot of dynamic range. It makes sense to me in a lot of cases!
@@CuivTheLazyGeek That was a very helpful response (and thank you very much). I'm had a feeling you could answer it. That said, while I do the research, can you explain why high ISO has "lower" read noise? And are we talking about upstream (pixel level) or downstream of the ISO amplifier to card? I guess I don't understand the mathematical formula (yet) on how with higher ISO there is less read noise and lower ISO high read noise. Take your time, but if I understand that piece to the puzzle, I think I can make it the rest of the way. BTW: Enjoyed the video a lot. Helped immensely.
Just a thought while we are doing min thoughts, will the noise % of a 1 second exposure not remain higher as each is being multiplied by the same factor. So the read noise will represent a much higher ratio than swamping the noise with longer exposures and a much higher signal. As a biologist it seems to me that you are extrapolating small sample data to equate with a much larger sample single sample which surely will include all data not present in the small data set?
@@CuivTheLazyGeek thank you for always replying to my questions . I really like your channel and think the content is amazing . I am learning a lot from you .
Cuiv, You make my brain hurt brother. Hehe. I’m sitting here star gazing and watching data roll in on M33 and I get this thought… “should I be taking subs longer or shorter than 180s?” So I pull up astrobin… lo and behold there’s a post today about 120s subs vs 300s Between that post and this video I’ve determined that I need to research this stuff when I’m not being nocturnal. Love your work man. CS (in Tokyo…. How could you even tell)
Cuiv, I just thought that I should let you know it's all your fault that I have pretty much switched from SGP to NINA!..... I thank you very much for that. :)
Hahaha, excellent to hear! NINA is an upgrade in so many respects (frame to frame overhead, AF, etc.) although it is somewhat of a downgrade for the sequencer as it is. I do hope it will help you capture great images!
Watched this series three years ago, now watch it again and clearly I feel I have more comprehension about this series. Good work!
That's awesome to hear!
This is one of the best explanations I’ve seen related to noise and optimal exposure length! It was so easy to follow that I can almost pretend my brain is 15 years younger. Long, dry, boring purely technical explanations of these complex concepts make me feel like I need to lie down and take a nap before my 4:00 PM early-bird special buffet dinner.
This was not one of those, and I appreciate it!
Glad this was useful!
Great stuff. Very different from many astronomy channels I follow. Nice to see some deeper explanations than just seeing unboxing and superficial review videos. Looking forward to see the next episodes.
Thank you! I like the small details of the hobby, even if I don't understand them all (far from it)...
@ Christian , i was so hopping some one would say that out loud...
Thanks again Cuiv, looking forward to seeing the rest of this series.
you are my favourite geek!
Excellent analogy!
Thank you! The bucket analogy is not new, but I think the surface tension as read noise is :)
oh...ay 20:08 you answered my question. Note to self...watch the whole video first
Excellent simplification of the topic. I wish my grad school imaging instructor had made that so easily understandable 32 years ago.
Thank you David - once I manage to build a time machine, I'll make sure to go back 32 years into the past and replace your instructor for the introduction to the topic ;) That's awesome though, did you end up working in the field you studied?
@@CuivTheLazyGeek I became a medical physicist, but instead of working in diagnostic imaging I work in therapeutic radiological physics, radiation oncology. I still use much of my knowledge of digital imaging in my work, but most of my passion for imaging is in my hobby. Thanks for producing these videos. You have a wonderful way of explaining things.
NINA made me laugh tonight. I was reading through the release notes and one entry stood out. Various grammar and SPELING errors corrected :)
With respect to my question on Celestron mount parking, it is looking like a bug in the Celestron driver as APT has the same problem.
Hahaha that is pretty cool! Almost wondering whether it has been inserted intentionally :) I should ask around. Bummer on the Celestron thing.... I hope it gets fixed - I used to have an account on their test/dev forum, and the devs were very reactive, so it may be worth asking.
Celestron has released a updated Unified Ascom driver.
Great explanation!
At some point I thought you could have made this video shorter and say the same, but I couldn't say it was boring. Thanks for making a quite complicated item clear with simple words.
Yep, I have a problem with conciseness but I do my best!
@@CuivTheLazyGeek that's totally okay!
Great explanation! Thank you very much, please keep going!
Thanks Kayed! Will definitely keep going!
Great educational post, I watched it twice!
Thank you for the feedback! It's always nice to know the details :)
So I have an off the wall question for you, but you might be able to help my understanding here. I haven't made it through this series of videos, but my question is a bit different. I understand the basics of "Shot" noise and "Read" noise (Upstream which can be amplified by dialing up amplification with ISO, and downstream - inherent to the camera). My question is this (sort of has to do with stacking - sort of). A well known photographer will "stack/combine" three 14-bit images in Photoshop (i.e. a portrait containing environment). He will photograph the scene using a strobe. He brackets his the three shots (not going to use the word exposure here) by only varying ISO. So flash power, aperture, and shutter speed remain constant. His goal is a 32-bit file in the largest color space to manipulate and print. His work is excellent, but I have a question regarding the science of his bracketing technique -- which he either can't or won't explain. In an ISO invariant sensor, is there any advantage of bracketing ISO two stops up and down or center vs taking the same image, copying it into three images, and combining those and then pushing the processing of highlights and shadows? I realize, that by copying the image twice, you eliminate the variance of environment "shot noise" differences, but do you think you can actually see the difference? It seems, by bracketing ISO, you're just introducing more noise then needed - certainly at the higher ISO. I do realize that some cameras (i.e. Nikon D850 has two processing engines that gives ISO 400 more DR than ISO 380). That aside, can you see any advantage to doing a bracket of ISO when keeping the other variables of SS/Aperture/flash power unchanged?
Well, ISO is just an amplifier. Higher ISO isn't intrinsically noisier than lower ISO in terms of SNR. In fact, higher ISOs typically have LOWER read noise (very important for astrophotography), but sacrifice dynamic range to some extent. Higher ISOs appear to have more noise, simply because normally higher ISO is associated with something like faster shutter speed (to freeze the moment for instance), but in that case everything, including the shot noise, gets amplified by the high ISO to achieve the same brightness level as a longer exposure with lower ISO. With a shorter exposure (regardless of ISO) you will achieve less signal, less shot noise (sqrt(signal)) and worse Signal to Shot Noise Ratio than a longer exposure. If you used higher ISO to achieve higher overall brightness you'll have LESS read noise, but also less dynamic range than an exposure with lower ISO.
With bracketing you have:
- an underexposed low ISO image with high dynamic range and high read noise (read noise is usually inconsequential in well lit conditions, but very important in astrophotography)
- a well exposed normal ISO image with good DR and "normal" read noise (sweet spot is usually around 800 for a lot of cameras in terms of read noise). Maybe the normally exposed image could have clipped highlights somewhere.
- an overexposed high ISO image with low dynamic range, potentially clipped highlights, low read noise.
Because all share the same parameters besides the ISO (which is just an amplifier, nothing more nothing less), they have sensibly the same SNR (low ISO slightly worse because of high read noise). But each covers a different dynamic range (say 14 stops for low iso image, 13 stops for medium iso image, 12 stops for high iso image), centered differently.
Depending on how the bracketed images are combined (as assume some kind of HDR replacement + averaging), you will have better SNR by increasing signal more than shot noise, because square root function (although the LOW ISO image does introduce slightly more read noise), AND an HDR picture that covers a lot of dynamic range.
It makes sense to me in a lot of cases!
@@CuivTheLazyGeek That was a very helpful response (and thank you very much). I'm had a feeling you could answer it. That said, while I do the research, can you explain why high ISO has "lower" read noise? And are we talking about upstream (pixel level) or downstream of the ISO amplifier to card? I guess I don't understand the mathematical formula (yet) on how with higher ISO there is less read noise and lower ISO high read noise. Take your time, but if I understand that piece to the puzzle, I think I can make it the rest of the way. BTW: Enjoyed the video a lot. Helped immensely.
looking for the best light rig setup based on the SvBony 503ed 80mm with the skywatcher adventurer GTI mount. Can you provide suggestions?
Wow, you must have made it rain indoors as you now really do have water in your "buckets" Magic......
I totally brought the buckets out in the rain first (not) :-)
Very well explained,good job man
Thank you! Those videos about technical details and PixInsight are much less popular, but I like making them as it's all valuable information.
very informative video! Thanks Cuiv :)
Just a thought while we are doing min thoughts, will the noise % of a 1 second exposure not remain higher as each is being multiplied by the same factor. So the read noise will represent a much higher ratio than swamping the noise with longer exposures and a much higher signal. As a biologist it seems to me that you are extrapolating small sample data to equate with a much larger sample single sample which surely will include all data not present in the small data set?
You are right, have a look at how that darn read noise stacks in the total stack here: ruclips.net/video/3RH93UvP358/видео.html
looking for the best light rig setup based on the SvBony 503ed 80mm. Can you provide suggestions?
Excellent video - thank you 🙏
May I ask - what type of noise is reduced with increased ISO ?
Shot noise ? Read noise ? Or somthing else ?
Read noise is reduced with increased ISO!
@@CuivTheLazyGeek thank you for always replying to my questions . I really like your channel and think the content is amazing . I am learning a lot from you .
Cuiv,
You make my brain hurt brother. Hehe.
I’m sitting here star gazing and watching data roll in on M33 and I get this thought… “should I be taking subs longer or shorter than 180s?”
So I pull up astrobin… lo and behold there’s a post today about 120s subs vs 300s
Between that post and this video I’ve determined that I need to research this stuff when I’m not being nocturnal.
Love your work man.
CS (in Tokyo…. How could you even tell)
Hahaha I know man, sorry lol - you can also check my recent videos (within the last couple of months) on the topic!
Cuiv, I just thought that I should let you know it's all your fault that I have pretty much switched from SGP to NINA!..... I thank you very much for that. :)
Hahaha, excellent to hear! NINA is an upgrade in so many respects (frame to frame overhead, AF, etc.) although it is somewhat of a downgrade for the sequencer as it is. I do hope it will help you capture great images!
Extremely useful! I still crack up at leeeeeters vs liters :)
Hahaha, at least my difficulties with English can entertain :D :D Glad this is useful!
Great!
Thank you!
Conveyer belt? Why not use a Sushi train. When in Rome.... Where would we be without randomness? Great video.
Oh no, I missed such a great opportunity!!! I hadn't even though of that :D Thank you!
DUDE this is sick
Now i know where the phrase "light bucket" came from in the 70s. (big dobs)
I had such a light bucket... But ended up not using it enough and said farewell... :)
@@CuivTheLazyGeek yea mine was a 16 inch F-4 it weighted a ton....................... LOL
After 8 minutes, 4 bowls and whole lot of spoken numbers, I stopped the video. This format did not work for me.