Sensor Analysis, Optimal Exposure Times, and LRGB Imaging Budget

[c172jeff]

Using Sharpcap and the information from this video...

Deep Sky Astrophotography With CMOS Cameras by Dr Robin Glover
https://youtu.be/3RH93UvP358

I have determined the optimal exposures time for my equipment as

Optimum Sub Exposure Length = in seconds
Bortle L RGB
6 5 16
5.5 8 25
5 13 40
4.5 22 66
4 37 110
3.5 46 141
3 59 176
2.5 64 196
2 70 207
1.5 80 235
1 88 271


Using NINA, I wish to build a profile for say a B2 sky location. I wish to only use L, R, G, and B filters

My table shows that the optimal exposure time for L is 70 seconds and the optimal exposure for the color filters are 207 seconds


I wish to use luminance layering/processing in pixinsight. I believe this means a minimal amount of color data and the detail arising from the most amount of Luminance images....

I wish to use NINA to sequence the exposures as L, followed by red, followed by green, then lastly by blue. I then plan to repeat as much as possible

If I rounded my exposure times of 70 and 207 to say 100 and 300, what should I do in NINA to collect the correct amount of image data...

I am thinking something like

Take 3 Luminance images at 300 seconds (maybe this 3 should be 9 instead)

Image 1 = 100 seconds of l
Image 2 = 100 seconds of L
Image 3 = 100 seconds of L

Then take the color

Image 4 = 300 seconds of red
Image 5 = 300 seconds of green
Image 6 = 300 seconds of blue

If this is one "loop" , these 6 images comprise 1200 (or 1800) seconds total of image data
If I do this as much as possible in one night and process...

I would think the best image would come just from using only the RGB and no luminance as this represents 900 seconds instead of 300 seconds of detail layered on 900 seconds of color....

Is this correct, how would one match the number of exposures to be taken for each filter stack to the optimal exposure times in my table..


Should I instead take 9 100 second luminance images per one cycle of r,g, and b?
Should I take even more Luminance detail images?

I believe after a handful of RGB sets (loops through an image sequence) where 900 seconds are collected, I can stop taking RGB completely and collect as much L as possible as that impacts the most detail in the context of diminishing returns of more color data.

This is a luminance layering question as well as what to make of optimal exposure times in the context of building an automated sequence. I realize one tends to want more L than RGB filter data, but the exposure times are smaller. How do I make it up with a proper number of exposures?

I was hoping someone can just say take x l images per a single r, g, and b image (given my exposure times)and explain why a little to get me thinking in the right direction

Thanks,
Jeff

[oopfan]

Hi Jeff,

It's fantastic to see that you have taken a scientific approach to imaging. I take a similar approach. I have a website that enables me to model my telescope, camera, and light pollution. Then, given that model I can apply it to pretty much any DSO for LRGB astrophotography. The range of DSO's include galaxies, planetary nebulae, and some reflection nebulae. Currently I am working on a model for emission nebulae. The brightness of a DSO is its Surface Brightness in magnitudes per square-arc-second. A lot of catalogs provide Surface Brightness but if a value is not available then you can calculate it with a simple formula for which I have a calculator. Large, irregular nebulae are problematic since they cannot be modeled with a simple ellipse. In these cases, one must rely on experience to estimate its surface brightness based on a test exposure. It is not as difficult as it sounds.

The website's code takes all of the inputs to calculate the DSO's photon flux per pixel, in addition to the effects of sky brightness, read noise, and dark noise. You tell it the exposure for a given filter, and it calculates the signal-to-noise ratio (SNR) per frame. Based on that SNR you may choose to increase or decrease the exposure. For example, I like to see SNR > 2 per frame. If it is less than that, I increase the exposure. Once you settle on exposure then you will adjust the frame count. This yields the SNR per stack. You repeat this for all four stacks: LRGB. The SNR of each stack contributes to the total SNR. I have discovered that Total SNR > 30 will likely win you an award an Astrobin, but in many cases it requires multiple sessions to achieve. On the other hand, SNR of 10 can produce satisfactory results.

One of the great benefits of this system is that it reduces waste by alerting you to weaknesses in your sensor. For example, my sensor is most sensitive to green but less sensitive to red and blue. Therefore, it tells me to spend less time capturing green frames and more time to red and blue. This is the underlying concept behind "color balance". The goal is to adjust the frame counts of color channels so that the SNR of the color stacks are approximately equal.

The last question is "How much color should I capture relative to luminance?" This is a hotly debated topic. Intuitively, we know that if we capture 100 luminance frames and only one frame each of color then that is not enough. After much debate we've reached a compromise: the SNR of each color channel should be 60% of the luminance SNR. So for example, if the luminance SNR is 10 then we should adjust the color frame counts such that red SNR is 6, green SNR is 6, and blue SNR is 6. Using the 60% rule ensures that the total SNR of the color stacks equals the SNR of the luminance stack.

Please PM me if you would like to find out more.

Brian

[rkymtnbiker]

Brian,

I'm glad to hear you discuss the brightness of our targets when considering exposures. I also watched Dr. Glover's presentations on which types of noise affect our SNR and how to calculate exposure times based upon sensor Read Noise and the Light Pollution in the Sky (along with Focal Ratio, OSC, and Filters used). But, those calculations don't take into account target brightness. I'd like to see how you calculate your exposure times based on SharpCap's Smart Histogram AND target brightness. Can you send me the link to your website?

Thanks,

Ted