Astrophotography Explained: Decoding Signal to Noise Ratio and Number of Sub Frames
Vložit
- čas přidán 20. 08. 2024
- I've been making videos about signal to noise ratio. I even have a free app that you can use to estimate it! But what does a good signal to noise ratio look like? The results may surprise you!
For those interested, here is the gear I use (or equivalent; Orion has discontinued some of their products, so the equivalent skywatcher products are listed). Most of the links are Amazon Affiliate links. Thanks!
From Amazon:
Skywatcher HEQ5 Mount amzn.to/3NVrVQi
Skywatcher 80mm F7.5 Doublet amzn.to/47mbfZi
ZWO EFW mini filter wheel amzn.to/48OxhoN
Optolong LRGB Filter Set (1.25") amzn.to/47r23Tw
Orion 0.8x Reducer amzn.to/47s64a4
USB RJ45 Cable for Controlling Mount amzn.to/4aPlIzu
Raspberry Pi to control mount amzn.to/48mTBWH
R Pi case with touch screen amzn.to/3HvdNKl
ZWO Autofocuser amzn.to/3Skg0hX
From Agena Astro:
ZWO 294MM Pro Camera: bit.ly/3tZmjOo
HEQ5 Mount: bit.ly/3O7VQVz (mine is Orion Sirius, which is equivalent)
ZWO Filter Wheel: bit.ly/3Soo5lw
Optolong Filters (LRGB, SII, Ha, OIII): bit.ly/3SlEPtJ (the Ha and OIII narrowband filters are better than what I have)
Skywatcher Evostar 80mm: bit.ly/4bfVjev (equivalent lens)
USB RJ45 Cable: bit.ly/48FRj50 (connects mount to laptop directly; mine is a different brand, but this should do)
ZWO Autofocuser: bit.ly/425XUDB
0.8x Reducer: bit.ly/3tQlbg7 (mine is Orion, but this should be equivalent)
Here is the link that compares Bortle 7 to Bortle 4: • Let's measure the SNR!...
I made a free SNR app that can help you know how much integration time you need. It's simple to use! Check it out here: deepskydetail....
#astronomy #astrophotography #neuralnetwork #gimp #rstudio #photography #machinelearning #free #celestron #zwo #siril
Drankin Song by Kevin MacLeod is licensed under a Creative Commons Attribution 4.0 license. creativecommon...
Try the sharpening tool I made (AstroSharp)! Download it from GitHub here:
github.com/dee...
nqFI29h8W0Fp9UkKOJry
Equipment Used or equivalent (Affiliate Links for Agena Astro and Amazon):
Skywatcher HEQ5 Mount amzn.to/3NVrVQi (mine is Orion, but this should be equivalent)
Skywatcher 80mm F7.5 Doublet amzn.to/47mbfZi (equivalent lens)
ZWO EFW mini filter wheel amzn.to/48OxhoN
Optolong LRGB Filter Set (1.25") amzn.to/47r23Tw
Orion 0.8x Reducer amzn.to/47s64a4
USB RJ45 Cable for Controlling Mount amzn.to/4aPlIzu
Raspberry Pi to control mount amzn.to/48mTBWH
R Pi case with touch screen amzn.to/3HvdNKl
ZWO Autofocuser amzn.to/3Skg0hX
ZWO 294MM Pro Camera: bit.ly/3tZmjOo
Skywatcher HEQ5 Mount: bit.ly/3O7VQVz (mine is Orion Sirius, which is equivalent)
ZWO Filter Wheel: bit.ly/3Soo5lw
Optolong Filters (LRGB, SII, Ha, OIII): bit.ly/3SlEPtJ (the Ha and OIII narrowband filters are better than what I have)
Skywatcher Evostar 80mm: bit.ly/4bfVjev (equivalent lens)
USB RJ45 Cable: bit.ly/48FRj50 (connects mount to laptop directly; mine is a different brand, but this should do)
ZWO Autofocuser: bit.ly/425XUDB
0.8x Reducer: bit.ly/3tQlbg7 (mine is Orion, but this should be equivalent)
One thing I learn is once you pass a certain amount of signal you don't pay attention to noise now u just start to see omg the structure is getting more and more define the more hours I put
I'm talking about enough signal were you don't need denoiser like over 100 hours
Yeah, I agree! You get to a certain point (given light pollution isn't a problem, e.g., you're using filters or at a dark site) where things just start to pop, and it's really cool when that happens :)
i agree with others, it’s very very helpful to actually see some good examples, it’s one thing to know the numbers but entirely different actually seeing the difference visually. great work, subscribed and i hope to see more!
Thank you!
Have to agree videos like this are really good I'm thankful for people making these especially since I'm trying to get into this hobby.
Your videos are exceptional, I'm thoroughly impressed with the amount of work you put into the animations. Admittedly I already have a pretty good grasp of SNR, but the animation work you are doing makes all of your topics you cover a million times easier to understand and reason about.
Thank you for your kind words! I enjoy trying to animate things, and it is awesome to know people find them useful, especially those more advanced in astrophotography and already have grasped the topic :)
This is so interesting and really clarifies some confusions I had I was expecting to see the number 100,000 next to your subscriber button but 2,000?! I can't wait until you blow up!
Thank you :)
So great !
That's really great stuff and a lot of work
Your channel is fantastic, keep on that great work !!!
Thank you! :)
Wow, great video. Easy enough for even someone like me to understand. I'm saving this one to watch again!
Glad you enjoyed it! Thank you!
30, 60 and 90 in decibels?
It's just a linear measurement, so no decibels. There are no units for the SNR measurements in the video as both the signal and noise are measured in the same way (e.g. ADU of the pixels). So it cancels out.
@@deepskydetail Ok. 👍
So do the stats really suggest that for a given period of time shorter subs will give a better result than longer ones. I've recently spent 220 mins on 3min subs, thinking that 73 subs should be well into the flat section of noise reduction by normal thinking. But 200 plus 1min sub would be better?
This contrasts with tribal knowledge of beyond 20 there's little benefit and sharp caps lectures which suggested to me ludicrous subs of 15mins, I'd have a white screen at that point, so currently aim for histogram around the first third.
Or is this because you are looking to SNR and stretch rather than an absolute level of signal? Perhaps computationally getting an absolute high signal was more important when processing hundreds of subs was tricky/not feasible?
My SNR app will give an estimation of what will happen if you change the sub exposure length. I think in general, it all depends on light pollution. If there's a lot of it, exposure length doesn't matter too much, just total integration time. If you're in dark skies, you might get slightly better results with longer subs. Hope that helps and I understood the question! :)
@@deepskydetailIt was written on the phone just before I fell asleep so might have been a bit garbled. But yes that's the gist of the question. I'm in bortle 4, and might have 2300 and 2200 as signal and background. Even with 70+ lights i'm not seeing a great SNR popping out of your app, but the gain is tricky to find for a canon RP, best estimates are 0.43 I think, and noise of 5.5. It's interesting in part because I'm looking, yet again, at M42 and with short subs I can get good detail on the core, so far I'm been trying short med and long subs so that I get the fainter dust away from the nebula itself, which leads to complexities in merging them etc. But I'm now thinking Just sit there are 10 or 15s subs for hours.
However there must be a minimum exposure? I mean the logical conclusion is 1s, but at that point most subs won't have any dust detail as you are below 1 photon per second hitting a sensor, and when most don't have anything hitting the sensor surely the stacked result is nothing for that pixel? So is there not a case for a minimum exposure time based on the likelihood of each pixel getting a true signal being > 50%, so your DSO + skyglow signal on a faint section should give that lower limit, no idea how?
@@stevejones7830 Yes, I think you are correct about minimum exposure! If you're not getting any photons on a sub frame for an area of the sky most of the time, then the average probably won't be high enough to offset the read noise of the camera. I'm not exactly sure what that minimum exposure would be though. I would bet that it is lower than we expect though, especially with a lot of light pollution! I think that as long as the histogram isn't cut off, you'd be good though :)
@@deepskydetail I'm sure you are already 1/2 way through this thought process. There will be a linear relationship between the brightness of the target and the photons/s (more likely /minute) So knowing sensitivities, magnitude etc. it might be possible to set a minimum time.