Another (but slightly different) question on total integration time and SNR - Forums

Topic starter

Jan 8, 2025

Hey all

so I’ve seen all the formulas and fully understand how stacking and total integration time works, but something that’s always baffled me is the following:

Firstly, if I take 3 hours of imaging with an OSC, is that the same as doing 1 hour each of RGB on mono, or do I effectively need to do 3 hours each of RGB on mono, so 9 hours in total to reach the same as 3 hours of a OSC?

and secondly, with SNR, (hopefully you can understand my question here) I know the formula for noise and stacking total frames, so if I take 10 frames of OSC, the SNR is easily calculated and similarly the formula can be applied to each mono filter, however, if I combine say SHO filters each shot with 10 frames, is that SNR formula now based off 30 frames or…?? Hopefully that makes sense!

bascially is SNR based off all the frames in each channel or each channel has is own SNR value? These are mostly mono questions as it’s simple and obvious with OSC

andrea tasselli

Frequent contributor

Jan 8, 2025

Paul:
Firstly, if I take 3 hours of imaging with an OSC, is that the same as doing 1 hour each of RGB on mono, or do I effectively need to do 3 hours each of RGB on mono, so 9 hours in total to reach the same as 3 hours of a OSC?

*It is neither one nor the other as spatial resolution and interpolation "noise" will affect the results but broadly speaking 3 hours of OSC are more than 1 hour each of RGB mono but less than 3 hours each of RGB mono, in terms of SNR. If I were to guess, I'd say the proportion is 1:2.25 for high end sensors. As for the second question the final SNR is based on the weighted SNR average of the final SHO composite, so in general proportional to the number of frames per channel.

Nooa Jutila

Jan 8, 2025 · Edited: Jan 8, 2025

This is a simplified explanation; read further for corrections

1 hour of RGB with a RGGB bayer pattern effectively has: 15 min R, 15 min x 2 G = 30min, 15 min B because 1/4 of the pixels have red filters, 2/4 have green and 1/4 have blue.

An OSC camera is a mono camera with a tiny filter for each pixel. With a dual bandpass filter, the blue channel receives very little signal, green has twice as many pixels and receives OIII well, and red has the same amount of pixels as blue but it gets a lot of Ha.
1 hour of RGB with a mono camera has 3 filters with each letting in a lot of light, with all the pixels recording data for a single color at a time.

For example, in terms of blue signal an OSC camera receives four times less signal than a mono camera with a blue filter.

You could roughly say that 1 hour of OSC data is equal to 1 hour of Mono data, but the mono data is 20 minutes for each channel and they have much better quality.

Helpful Insightful

Wei-Hao Wang

Jan 8, 2025

It's good to let the data speak for themselves. Check this out:

https://www.cloudynights.com/topic/858009-cooled-mono-astro-camera-vs-modified-dslrmirrorless/?p=12402843

Same Sony CMOS. One cooled mono. One OSC (modified mirrorless) and uncooled. Same optics. Same exposure time. Similar conditions. Similarly processed. Can you tell which one is which?

Engaging

Tony Gondola

Frequent contributor

Jan 8, 2025

You can't really say that in OSC you any get 1/4 of the Ha signal because Bayer Matrix filters aren't perfect. There is a certain amount of leakage across all the channels and it varies with wavelength. In the near IR wavelengths that a lot of CMOS cameras are sensitive to, the matrix is almost transparent. This is good and bad, you get more total signal but none of your narrowband channels are pure. That's why the difference between MONO and OSC isn't as great as some might think.

Helpful Insightful Concise

D. Jung

Jan 8, 2025 · Edited: Jan 8, 2025

Using an osc
- exposing 1/4 of the sensor doesn't mean you get 1/4 of the signal or SNR, the debayer process reconstructs the full image. Those pixels get the same amount of light minus some losses due to the bayer filter
- the debayer process is noisy and introduces different types of noise that you don't get with mono camera. It's difficult to put a number on that but it's not negligible
- up and foremost effect is that you significantly reduce the spatial resolution of your imaging system. For red and blue, your Nyquist frequency drops by more than a factor of 2. The worst loss of resolution is along the diagonals. That effect is very visible in fine structures

I switched to mono after using osc for many years. My conclusion, I wish I had switched to mono years ago

Just to show what the debayer process does. Take an image and run the debayer VNGprocess over it (this algorithm is not using information from other color channels so it should be a fair representation when using on a color image). Left is original and right is after debayer.

Helpful Insightful Engaging Supportive

Ashraf AbuSara

Jan 8, 2025 · Edited: Jan 8, 2025

Wei-Hao Wang:
It's good to let the data speak for themselves. Check this out:

https://www.cloudynights.com/topic/858009-cooled-mono-astro-camera-vs-modified-dslrmirrorless/?p=12402843

Same Sony CMOS. One cooled mono. One OSC (modified mirrorless) and uncooled. Same optics. Same exposure time. Similar conditions. Similarly processed. Can you tell which one is which?

I am assuming the mono is the one to the right? It looks significantly better to my eyes. But to be fair the OSC is an uncooled chip, so some of that difference maybe bridged with a cooled OSC.

I also would caution against drawing conclusions from one target. Using a bright target will get you decent signal no matter which sensor you use. But if you look towards the dark /fainter features in your image, it looks quite a bit cleaner and more well defined in your mono image. Was that comparison with LRGB or just pure RGB?

That being said the stated difference between Mono and OSC is exaggerated a lot and OSC can achieve great results with similar times.

Helpful Insightful Respectful Engaging

Wei-Hao Wang

Jan 8, 2025

Ashraf AbuSara:
Wei-Hao Wang:
It's good to let the data speak for themselves. Check this out:

https://www.cloudynights.com/topic/858009-cooled-mono-astro-camera-vs-modified-dslrmirrorless/?p=12402843

Same Sony CMOS. One cooled mono. One OSC (modified mirrorless) and uncooled. Same optics. Same exposure time. Similar conditions. Similarly processed. Can you tell which one is which?

I am assuming the mono is the one to the right? It looks significantly better to my eyes. But to be fair the OSC is an uncooled chip, so some of that difference maybe bridged with a cooled OSC.

I also would caution against drawing conclusions from one target. Using a bright target will get you decent signal no matter which sensor you use. But if you look towards the dark /fainter features in your image, it looks quite a bit cleaner and more well defined in your mono image.

Was that comparison with LRGB or just pure RGB?

I don't want to spoil it here. There are answers to all your questions in that thread.

Well Written

Rick Krejci

Jan 8, 2025

Ashraf AbuSara:
I am assuming the mono is the one to the right? It looks significantly better to my eyes. But to be fair the OSC is an uncooled chip, so some of that difference maybe bridged with a cooled OSC.

I also would caution against drawing conclusions from one target. Using a bright target will get you decent signal no matter which sensor you use. But if you look towards the dark /fainter features in your image, it looks quite a bit cleaner and more well defined in your mono image.

Was that comparison with LRGB or just pure RGB?

From #13 on that thread: "The left one was taken with Moravian C5A-100M. The right one was taken with Fujifilm GFX-100S"

Well Written Respectful

Wei-Hao Wang

Jan 8, 2025

D. Jung:
Using an osc
- exposing 1/4 of the sensor doesn't mean you get 1/4 of the signal or SNR, the debayer process reconstructs the full image. Those pixels get the same amount of light minus some losses due to the bayer filter
- the debayer process is noisy and introduces different types of noise that you don't get with mono camera. It's difficult to put a number on that but it's not negligible
- up and foremost effect is that you significantly reduce the spatial resolution of your imaging system. For red and blue, your Nyquist frequency drops by more than a factor of 2. The worst loss of resolution is along the diagonals. That effect is very visible in fine structures

I switched to mono after using osc for many years. My conclusion, I wish I had switched to mono years ago

Just to show what the debayer process does. Take an image and run the debayer VNGprocess over it (this algorithm is not using information from other color channels so it should be a fair representation when using on a color image). Left is original and right is after debayer.

If you use PI, and if your Bayer array under-sample the image, the most logical solution is to employ Bayer drizzle for registration and stacking, instead of using any debayer method, including VNG. If done properly, Bayer drizzle should give you a result identical to mono sensors of the same pixel size. Of course, you need to have more than a handful of dithered exposures to start with. If you are talking about single exposure, then it's true that mono can be better if the pixel size is too big compared to the star FWHM.

Helpful Insightful Concise

Arun H

Frequent contributor

Jan 8, 2025

Wei-Hao Wang:
Same Sony CMOS. One cooled mono. One OSC (modified mirrorless) and uncooled. Same optics. Same exposure time. Similar conditions. Similarly processed. Can you tell which one is which?

Hi Wei-Hao, thank you for sharing that comparison. Indeed, it would make sense that Bayer drizzled OSC would give near identical results to mono RGB. In fact, the OSC sensor being more biased towards the green (RGGB) actually probably makes it a bit closer to LRGB mono than pure RGB mono for the same integration time since the luminance data usually strengthens the green channel more than the Blue and Red when combined with RGB.

I do think one consideration is imaging under light pollution. I find that it is easier to handle gradients with mono RGB than with OSC since the LP from one channel does not affect the others due to the sharper cutoffs with the mono RGB filters. But under dark skies, it is quite easy and convenient to use OSC to get very good results.

Well Written Helpful Insightful Respectful Engaging

Ashraf AbuSara

Jan 8, 2025 · Edited: Jan 8, 2025

Wei-Hao Wang:
Ashraf AbuSara:
Wei-Hao Wang:
It's good to let the data speak for themselves. Check this out:

https://www.cloudynights.com/topic/858009-cooled-mono-astro-camera-vs-modified-dslrmirrorless/?p=12402843

Same Sony CMOS. One cooled mono. One OSC (modified mirrorless) and uncooled. Same optics. Same exposure time. Similar conditions. Similarly processed. Can you tell which one is which?

I am assuming the mono is the one to the right? It looks significantly better to my eyes. But to be fair the OSC is an uncooled chip, so some of that difference maybe bridged with a cooled OSC.

I also would caution against drawing conclusions from one target. Using a bright target will get you decent signal no matter which sensor you use. But if you look towards the dark /fainter features in your image, it looks quite a bit cleaner and more well defined in your mono image.

Was that comparison with LRGB or just pure RGB?

I don't want to spoil it here. There are answers to all your questions in that thread.

Interesting! I would have never guessed but the cooled RGB Mono looks far worse to me in that comparison.

Wei-Hao Wang

Jan 9, 2025

Ashraf AbuSara:
Interesting! I would have never guessed but the cooled RGB Mono looks far worse to me in that comparison.

Hi Ashraf,

Actually I wouldn't say the mono one looks any worse than the OSC one. It's just that their noise properties are different.

In this comparison, I did not employ Bayer drizzle for the stacking of the OSC images. This is because the seeing (and wind) is bad enough that the Bayer array very well sampled the stellar FWHM, and therefore drizzle does not bring any advantages on resolution. Without Bayer drizzle, noise amplitude is suppressed by debayerization, and has a larger correlation length among pixels. For eyes that only look at the amplitude and not the correlation, it's very easy to form an impression that this image has less noise. But it is not true.

Equivalently, on the mono image, one can run a noise reduction (or a simple Gaussian smoothing of small radius, small enough to not damage resolution at an apparent level) to suppress the noise amplitude and broaden the correlation length. This will make the mono image much more like (or even identical to) the OSC one. So, one way to look at this comparison is that the two images intrinsically have very similar amount of noise (or even identical amount) to start with, but one had gone through more smoothing because of debayerization to get the less noisy look while the other had not.

In a different CN thread, one very experienced OSC user correctly pointed out which is which, and with correct reasons (so it's not a lucky guess).

Well Written Helpful Insightful Engaging