Choose camera on FWHM?

[MilwaukeeLion]

Curious if anyone chooses which camera to image with based on evening seeing?  Was looking at different camera combo resolutions paired with ED80 at ok seeing FWHM of 2-4". Seems odd that ASI120mm is better option than DSLR 4.88 um with ok seeing.  Also seems odd if seeing improved during night both become worse - significantly under sampled.  Should be rooting for poor seeing on nights I want to use my ED80 and DSLR?

[vlaiv]

Maybe you should use different approach given your choice of sensors?

Rather than thinking in terms of maximum resolution on a given night for particular target, you decide on target based on conditions for any given night?

ASI120mm is not really suitable for large targets unless you want to do mosaic. DSLR on the other hand is very good for wider field and larger targets. Less than optimal seeing - go for something large enough that won't suffer so much because of missing finest detail. On a good night choose to do high resolution smaller targets - like PNs or small galaxies?

[MilwaukeeLion]

12 hours ago, vlaiv said:

Maybe you should use different approach given your choice of sensors?

Rather than thinking in terms of maximum resolution on a given night for particular target, you decide on target based on conditions for any given night?

ASI120mm is not really suitable for large targets unless you want to do mosaic. DSLR on the other hand is very good for wider field and larger targets. Less than optimal seeing - go for something large enough that won't suffer so much because of missing finest detail. On a good night choose to do high resolution smaller targets - like PNs or small galaxies?

Just seems counter intuitive to me being I mostly do lucky imaging / look at critical sampling and complain seeing was poor using C11.  With ED80 using same cameras with exceptional seeing you get significant under sampling vs just ok seeing which is ideal pixel size.  More excuses for poor result, can never have enough of those.

[vlaiv]

Just now, MilwaukeeLion said:

Just seems counter intuitive to me being I mostly do lucky imaging / look at critical sampling and complain seeing was poor using C11.  With ED80 using same cameras with exceptional seeing you get significant under sampling vs just ok seeing which is ideal pixel size.  More excuses for poor result, can never have enough of those.

You don't get that much under sampling as you might think. Actually with 80mm scope, I don't recommend going below say 1.5"/px - 2"/px because there is no point really.

There is quite a bit of difference in optimal sampling rates of DSO AP vs planetary AP.

With lucky imaging you are used to going after critical resolution - maximum provided by aperture. You discard lesser frames and in the end you have enough SNR to do some sort of frequency restoration process - be that wavelets or deconvolution or whatever.

With DSO ap things are quite different - there is different metric that is associated with optimum sampling rate and it is FWHM of stars in your subs. Scope aperture comes into that equation, but there are other contributing factors - seeing, guiding/tracking precision and pixel blur (that one is important to some extent in planetary ap as well).

We are used to think about seeing in terms of FWHM, but most of the time, FWHM that we measure, even in short exposure like 2s FWHM where guiding/tracking error is minimized, still depends on aperture as well as seeing conditions. You will get larger FWHM with smaller scope for the same atmospheric seeing.

Actual seeing (instrument independent one) is actually better in terms of FWHM than people usually suspect. It is often in range 1-2". Local seeing can be quite different to that if there are local thermals as you undoubtedly know (scope thermals, local heat sources like concrete / asphalt pavements and roads, houses even bodies of water). These have cumulative effect in long exposure.

In any case, here are some "tips" for sampling rate when doing DSO imaging:

- On most nights with 80ED you will have about 3"-4" FWHM, not because of seeing, but because of all things combined. Different scope will have smaller FWHM than that (for example 8" reflector)

- Let's say optimum sampling rate (but things are not so clear cut here as in planetary, because we use Gaussian approximation, and gaussian shape is not band limited so we can't find critical frequency when doing such approximation - but we can find sensible cut off point) is FWHM / 1.6. For example for 3" FWHM you would need to sample at 1.875"/px, while for 4" that would need to be about 2.5"/px.

- Due to pixel blur you can in practice go as low as x2-3 that value, meaning you can go up to 4-6"/px and you won't get aliasing artifacts from under sampling. Blocky stars are sort of a myth. You can get single pixel stars if you under sample greatly, but it is up to interpolation algorithm when you "zoom in" (and in reality you should not zoom in past 1:1 - meaning one image pixel per one device pixel) how that star will look. Nearest neighbor interpolation will indeed make it look like square, but other interpolation algorithms will not (they might have other "artifacts"). After all - just look at images done with simple lens of short focal length (like samyang 135 or even 50mm lens) - no blocky stars, although aperture is comparable to 80mm (Samyang 135 is F/2 lens so aperture is 67.5mm - very close to 80mm scope).

- DSLR or any other OSC sensor will sample at twice "base" pixel size in reality - due to bayer matrix, every other pixel "counts" as sampling point for particular color, so samples are spaced not single pixel width but double of that. However pixel blur will be that of single pixel width

 

 

[MilwaukeeLion]

2 hours ago, vlaiv said:

You don't get that much under sampling as you might think. Actually with 80mm scope, I don't recommend going below say 1.5"/px - 2"/px because there is no point really.

There is quite a bit of difference in optimal sampling rates of DSO AP vs planetary AP.

With lucky imaging you are used to going after critical resolution - maximum provided by aperture. You discard lesser frames and in the end you have enough SNR to do some sort of frequency restoration process - be that wavelets or deconvolution or whatever.

With DSO ap things are quite different - there is different metric that is associated with optimum sampling rate and it is FWHM of stars in your subs. Scope aperture comes into that equation, but there are other contributing factors - seeing, guiding/tracking precision and pixel blur (that one is important to some extent in planetary ap as well).

We are used to think about seeing in terms of FWHM, but most of the time, FWHM that we measure, even in short exposure like 2s FWHM where guiding/tracking error is minimized, still depends on aperture as well as seeing conditions. You will get larger FWHM with smaller scope for the same atmospheric seeing.

Actual seeing (instrument independent one) is actually better in terms of FWHM than people usually suspect. It is often in range 1-2". Local seeing can be quite different to that if there are local thermals as you undoubtedly know (scope thermals, local heat sources like concrete / asphalt pavements and roads, houses even bodies of water). These have cumulative effect in long exposure.

In any case, here are some "tips" for sampling rate when doing DSO imaging:

- On most nights with 80ED you will have about 3"-4" FWHM, not because of seeing, but because of all things combined. Different scope will have smaller FWHM than that (for example 8" reflector)

- Let's say optimum sampling rate (but things are not so clear cut here as in planetary, because we use Gaussian approximation, and gaussian shape is not band limited so we can't find critical frequency when doing such approximation - but we can find sensible cut off point) is FWHM / 1.6. For example for 3" FWHM you would need to sample at 1.875"/px, while for 4" that would need to be about 2.5"/px.

- Due to pixel blur you can in practice go as low as x2-3 that value, meaning you can go up to 4-6"/px and you won't get aliasing artifacts from under sampling. Blocky stars are sort of a myth. You can get single pixel stars if you under sample greatly, but it is up to interpolation algorithm when you "zoom in" (and in reality you should not zoom in past 1:1 - meaning one image pixel per one device pixel) how that star will look. Nearest neighbor interpolation will indeed make it look like square, but other interpolation algorithms will not (they might have other "artifacts"). After all - just look at images done with simple lens of short focal length (like samyang 135 or even 50mm lens) - no blocky stars, although aperture is comparable to 80mm (Samyang 135 is F/2 lens so aperture is 67.5mm - very close to 80mm scope).

- DSLR or any other OSC sensor will sample at twice "base" pixel size in reality - due to bayer matrix, every other pixel "counts" as sampling point for particular color, so samples are spaced not single pixel width but double of that. However pixel blur will be that of single pixel width

 

 

Good stuff, adding to your folder of critical sampling info and formula to calculate size of balloon passing in front of sun.