Advice requested for 12 panel mosaic

[powerlord]

Hi chaps, looks like awesome seeing forecast for tonight.

I thought I'd go for a 12 panel mosaic with my C9.25 with 2x barlow and asi1600.

First question: What filter ? I'm thinking either the zwo Ha or the cheap svbony IR pass I have ? Im not sure how good or bad the ir pass is though tbh.. whereas I know my Ha is nice from my DSO work. So.. Ha ?

Second question: Gain wise, I always use unity 139 for DSO, but I'll be shooting 10 bit video here, and I assume I want gain much higher. is there a best setting ? and in firecapture do I need to set offsets ? (in asiair it takes care of this so never had to bother with it)

For each panel, my plan was to shot full frame 4656x3520 in 10 bit for 8 mins:  should get me 23fps according to spec (will try later this afternoon) - so 8 mins should get me >10k frames. that seem enough ?

For moving rig, I think I have to do that manually, using firecapure liveview and using asiair controls to move mount.. which is gonna be a pain, but Im on a mac for capture and it seems firecapture doesn't do mosaics ? so MoonPanorama maker not an option.

So it's a case of setting up, capturing with a 10k limit on frames, move to next panel, repeat I think ?

any other tips ?

[neil phillips]

16 minutes ago, powerlord said:

Hi chaps, looks like awesome seeing forecast for tonight.

I thought I'd go for a 12 panel mosaic with my C9.25 with 2x barlow and asi1600.

First question: What filter ? I'm thinking either the zwo Ha or the cheap svbony IR pass I have ? Im not sure how good or bad the ir pass is though tbh.. whereas I know my Ha is nice from my DSO work. So.. Ha ?

Second question: Gain wise, I always use unity 139 for DSO, but I'll be shooting 10 bit video here, and I assume I want gain much higher. is there a best setting ? and in firecapture do I need to set offsets ? (in asiair it takes care of this so never had to bother with it)

For each panel, my plan was to shot full frame 4656x3520 in 10 bit for 8 mins:  should get me 23fps according to spec (will try later this afternoon) - so 8 mins should get me >10k frames. that seem enough ?

For moving rig, I think I have to do that manually, using firecapure liveview and using asiair controls to move mount.. which is gonna be a pain, but Im on a mac for capture and it seems firecapture doesn't do mosaics ? so MoonPanorama maker not an option.

So it's a case of setting up, capturing with a 10k limit on frames, move to next panel, repeat I think ?

any other tips ?

I often prefer IR685 works well for me. Dont like giving advice in case it doesnt work out. But i will say. I dont like high gain on the moon it makes the images grainy. I know about read noise. But hi gain on the moon also produces noise. 

Just dont let gain noise dominate. 

[vlaiv]

I'd reconsider whole ordeal.

First, ASI1600 has 3.8µm pixel size and you plan on using Ha filter with it. For that combination ideal F/ratio is F/11.6. You'll be much closer to it without the barlow then with. With barlow you'll be very oversampled.

Second problem that I see is off axis aberrations. I don't really know how much coma SCT produces. I think I read somewhere that coma in SCT is comparable to same aperture F/8 newtonian (not sure about that - but let's go with it).

Coma in newtonian is function of distance from optical axis and diffraction limited field is given by h = F^3/90 where F is F/ratio and h is distance from optical axis in mm. F/8 newtonian will have diffraction limited field of 5.6888 = ~5.7mm. That will be radius of diffraction limited field.

Diameter will be about 11.4mm. Diagonal of ASI1600 is 21mm, this means that only inner 1/4 of the field will be diffraction limited and the rest will be distorted.

You can actually see this in SCT DSO images:

Center field stars with C9.25 and ASI1600:

Top right corner:

mind you - this is scaled down to about 1/4 to get meaningful resolution for DSO, true star shape will be:

(again - stars are very big due to long exposure and seeing effect - but you can see that coma extends several pixels from the center of the star).

In the end, here are the settings I'd recommend:

1. Use ROI of 1920×1680 px at 12 bit capture (16 bit SER)

2. Use 261 gain

3. Limit your exposure to 5-6ms regardless how faint image looks like and what histogram says

4. Do get 10000+ frames - that means 6-7 minute capture per panel (30fps at 1920x1680 at 16 bit)

5. You'll need more panels to cover whole lunar disk, but if you are not capturing full moon - you might need less than number that covers whole disc because there is no point in capturing beyond terminator - it will just be black like surrounding space

46 minutes ago, neil phillips said:

I dont like high gain on the moon it makes the images grainy. I know about read noise. But hi gain on the moon also produces noise. 

Just dont let gain noise dominate. 

If you fix your exposure length - and one should, then shot noise is fixed by amount of signal one gets. Changing gain will not change SNR in that regard. Higher gain will just lower read noise so overall SNR per sub will be better.

[Yawning Angel]

1 hour ago, powerlord said:

any other tips ?

Choose your path first, and stick to it. A 'Z' path has less slewing, and is faster but I find I can forget which panel is next if I'm distracted for a moment. An 'E' path mean the next from is always to the right 🙂

Allow plenty of overlap, it's easy to forget the vertical and get fixated on the horizontal when doing it manually.

Make sure your exposure take in account the awfully bright areas, which can sneak up on you, 8 panels in 🙂 

Try your green filter, you might find it sharper if the seeing is favourable

Never try and do this with a full RGB pass - your sanity won't survive!

[powerlord]

ok, thanks for that. how about using the x6.3 reducer/corrector and then adding the barlow ? or is that madness ?

[vlaiv]

6 minutes ago, powerlord said:

ok, thanks for that. how about using the x6.3 reducer/corrector and then adding the barlow ? or is that madness ?

I would not do that.

It might fix edge stars but it will most certainly render scope non diffraction limited on axis. All correctors / flatteners do that. They were made for long exposure imaging where atmosphere has much greater impact on image sharpness so one can afford not to be diffraction limited.

With lucky imaging - well you want to best sharpness instrument can provide.

Here - look at F/6.3 reducer paired with C11:

In particular - look at "AXIS" part in the bottom middle section.

Spot diagrams are larger than airy disk and Strehl is reduced to 0.695 even on axis

[powerlord]

okdokey - will it all a try then tonight. fingers crossed.

I'm breaking my well learned rule and trying too many new things tonight - new second mount, first time with 80ed and flattener on that, etc...

[neil phillips]

17 hours ago, vlaiv said:

I'd reconsider whole ordeal.

First, ASI1600 has 3.8µm pixel size and you plan on using Ha filter with it. For that combination ideal F/ratio is F/11.6. You'll be much closer to it without the barlow then with. With barlow you'll be very oversampled.

Second problem that I see is off axis aberrations. I don't really know how much coma SCT produces. I think I read somewhere that coma in SCT is comparable to same aperture F/8 newtonian (not sure about that - but let's go with it).

Coma in newtonian is function of distance from optical axis and diffraction limited field is given by h = F^3/90 where F is F/ratio and h is distance from optical axis in mm. F/8 newtonian will have diffraction limited field of 5.6888 = ~5.7mm. That will be radius of diffraction limited field.

Diameter will be about 11.4mm. Diagonal of ASI1600 is 21mm, this means that only inner 1/4 of the field will be diffraction limited and the rest will be distorted.

You can actually see this in SCT DSO images:

Center field stars with C9.25 and ASI1600:

Top right corner:

mind you - this is scaled down to about 1/4 to get meaningful resolution for DSO, true star shape will be:

(again - stars are very big due to long exposure and seeing effect - but you can see that coma extends several pixels from the center of the star).

In the end, here are the settings I'd recommend:

1. Use ROI of 1920×1680 px at 12 bit capture (16 bit SER)

2. Use 261 gain

3. Limit your exposure to 5-6ms regardless how faint image looks like and what histogram says

4. Do get 10000+ frames - that means 6-7 minute capture per panel (30fps at 1920x1680 at 16 bit)

5. You'll need more panels to cover whole lunar disk, but if you are not capturing full moon - you might need less than number that covers whole disc because there is no point in capturing beyond terminator - it will just be black like surrounding space

If you fix your exposure length - and one should, then shot noise is fixed by amount of signal one gets. Changing gain will not change SNR in that regard. Higher gain will just lower read noise so overall SNR per sub will be better.

I think i understand the effects of high gain after imaging for 15 years Vlaiv. High gain produces noise too.  Sometimes exposure is effected by transparency in situations like that obviously it is better to use a slower exposure than impossibly high gain. Unless one wants to use increasingly large stacks.  If that was your point then agreed. Otherwise i will continue to use what my eyes see when i capture and stack. I should have spotted some effects by now after imaging for 15 years. I have countless examples of gain being too high. Because exposure was too fast. Obviously its good to have a fast exposure. But not with gain so high that all you see is grain. A balance must be struck. Are you suggesting that gain can not be too high ? i was not talking about the relationship between exposure and gain. Just a comment on not letting gain get too high. 

Edited January 13, 2022 by neil phillips

[Freddie]

19 hours ago, vlaiv said:

I'd reconsider whole ordeal.

First, ASI1600 has 3.8µm pixel size and you plan on using Ha filter with it. For that combination ideal F/ratio is F/11.6. You'll be much closer to it without the barlow then with. With barlow you'll be very oversampled.

Second problem that I see is off axis aberrations. I don't really know how much coma SCT produces. I think I read somewhere that coma in SCT is comparable to same aperture F/8 newtonian (not sure about that - but let's go with it).

Coma in newtonian is function of distance from optical axis and diffraction limited field is given by h = F^3/90 where F is F/ratio and h is distance from optical axis in mm. F/8 newtonian will have diffraction limited field of 5.6888 = ~5.7mm. That will be radius of diffraction limited field.

Diameter will be about 11.4mm. Diagonal of ASI1600 is 21mm, this means that only inner 1/4 of the field will be diffraction limited and the rest will be distorted.

You can actually see this in SCT DSO images:

Center field stars with C9.25 and ASI1600:

Top right corner:

mind you - this is scaled down to about 1/4 to get meaningful resolution for DSO, true star shape will be:

(again - stars are very big due to long exposure and seeing effect - but you can see that coma extends several pixels from the center of the star).

In the end, here are the settings I'd recommend:

1. Use ROI of 1920×1680 px at 12 bit capture (16 bit SER)

2. Use 261 gain

3. Limit your exposure to 5-6ms regardless how faint image looks like and what histogram says

4. Do get 10000+ frames - that means 6-7 minute capture per panel (30fps at 1920x1680 at 16 bit)

5. You'll need more panels to cover whole lunar disk, but if you are not capturing full moon - you might need less than number that covers whole disc because there is no point in capturing beyond terminator - it will just be black like surrounding space

If you fix your exposure length - and one should, then shot noise is fixed by amount of signal one gets. Changing gain will not change SNR in that regard. Higher gain will just lower read noise so overall SNR per sub will be better.

How many of those 10,000+ frames would you propose are stacked?

[vlaiv]

6 hours ago, neil phillips said:

I think i understand the effects of high gain after imaging for 15 years Vlaiv. High gain produces noise too.  Sometimes exposure is effected by transparency in situations like that obviously it is better to use a slower exposure than impossibly high gain. Unless one wants to use increasingly large stacks.  If that was your point then agreed. Otherwise i will continue to use what my eyes see when i capture and stack. I should have spotted some effects by now after imaging for 15 years. I have countless examples of gain being too high. Because exposure was too fast. Obviously its good to have a fast exposure. But not with gain so high that all you see is grain. A balance must be struck. Are you suggesting that gain can not be too high ? i was not talking about the relationship between exposure and gain. Just a comment on not letting gain get too high. 

I'm saying that, once you dial in your exposure length - one that is short enough to freeze the seeing but not shorter than that, you can raise the gain as much as you want until you hit saturation / clipping and it won't change SNR.

In fact, if it were not for gain/read noise relationship - it would not matter which gain was used as gain is only conversion factor. For any given S/N if you use multiplicative constant - you don't change S/N as it is equal to S*c / N*c - constant "cancels out".

We recommend high gain simply because of this:

It is better to have 0.8e read noise than it is to have 1.5e read noise. In long exposure AP - it does not really matter what the read noise is - we can lessen its impact on final stack by choosing sub exposure length. With planetary we don't have that luxury as we are limited with seeing - we can't use arbitrary long subs. For this reason we want as low read noise as possible.

Just to reiterate - read noise is only thing that makes difference between stack of short and long subs (or even single very long sub) that add up to same total time - in terms of SNR. If we had zero read noise, then SNR would be the same for same total time - regardless of individual sub duration.

Eyes are not very good judge of SNR - look at this:

It might look like right side is much worse as far as noise is concerned - much more grain, but in reality - they have exact same SNR, except right side has been made brighter by multiplying the data by 3.

[vlaiv]

2 hours ago, Freddie said:

How many of those 10,000+ frames would you propose are stacked?

That really depends on seeing on particular night. Start with 5% and if seeing is good - increase number.

Autostakkert!3 gives nice "seeing / sub quality" graph - in time and sorted - and you can judge based on that graph how much subs to keep - but even if you keep only 5% - that is 400 subs out of 10000 and that improves SNR by factor of x20. Luna is rather bright and individual subs should already have decent SNR.

[neil phillips]

9 minutes ago, vlaiv said:

I'm saying that, once you dial in your exposure length - one that is short enough to freeze the seeing but not shorter than that, you can raise the gain as much as you want until you hit saturation / clipping and it won't change SNR.

In fact, if it were not for gain/read noise relationship - it would not matter which gain was used as gain is only conversion factor. For any given S/N if you use multiplicative constant - you don't change S/N as it is equal to S*c / N*c - constant "cancels out".

We recommend high gain simply because of this:

It is better to have 0.8e read noise than it is to have 1.5e read noise. In long exposure AP - it does not really matter what the read noise is - we can lessen its impact on final stack by choosing sub exposure length. With planetary we don't have that luxury as we are limited with seeing - we can't use arbitrary long subs. For this reason we want as low read noise as possible.

Just to reiterate - read noise is only thing that makes difference between stack of short and long subs (or even single very long sub) that add up to same total time - in terms of SNR. If we had zero read noise, then SNR would be the same for same total time - regardless of individual sub duration.

Eyes are not very good judge of SNR - look at this:

It might look like right side is much worse as far as noise is concerned - much more grain, but in reality - they have exact same SNR, except right side has been made brighter by multiplying the data by 3.

Well i think my eyes are infact a perfect judge, of what settings are doing in the feild. Theory is fine in practice. But infact the results can often look very different to the theories you keep stating. if grain is brighter it may infact be influencing the noise level i see when sharpening. What ever the reason. If its more obviouse. it matters not if theory says its the same ? i do see its effect. for you to say i dont is pretty bizzare. We may have to agree to disagree. 

[vlaiv]

11 minutes ago, neil phillips said:

Well i think my eyes are infact a perfect judge, of what settings are doing in the feild. Theory is fine in practice. But infact the results can often look very different to the theories you keep stating. if grain is brighter it may infact be influencing the noise level i see when sharpening. What ever the reason. If its more obviouse. it matters not if theory says its the same ? i do see its effect. for you to say i dont is pretty bizzare. We may have to agree to disagree. 

Of course, I could be wrong as a next person. I meant no disrespect by strongly advocating my point of view.

[vlaiv]

@neil phillips

May I propose simple experiment?

It consists of taking two flats with your scope and camera. Either with flat panel if you own one or just white wall (but make sure lighting conditions don't change). Exact focus is not critical as in flats - we just need means to get two simple to measure recordings of light.

Do one shot with low gain and one with high gain, make sure you have all other settings the same - exposure time, etc ...

Scale high gain image with ratio of mean ADUs of both images (so that they have same mean ADU) and compare standard deviation of both shots. One with higher standard deviation is more noisy (signal will be equal if mean ADU is equal).

[neil phillips]

37 minutes ago, vlaiv said:

@neil phillips

May I propose simple experiment?

It consists of taking two flats with your scope and camera. Either with flat panel if you own one or just white wall (but make sure lighting conditions don't change). Exact focus is not critical as in flats - we just need means to get two simple to measure recordings of light.

Do one shot with low gain and one with high gain, make sure you have all other settings the same - exposure time, etc ...

Scale high gain image with ratio of mean ADUs of both images (so that they have same mean ADU) and compare standard deviation of both shots. One with higher standard deviation is more noisy (signal will be equal if mean ADU is equal).

Bit busy processing and out capturing again tonight. But at some point i will capture some lunar and keep all settings the same. except one shot i will use high gain. and post results up here. I cant do it for a while its not something i can be bothered to think about to be honest. Because i know the effect high gain has on my sharpening routines. Have no real need to prove it. But if i can be bothered and have time at some point in the future. i will run some examples.

Untill then. Back to my processing take care 

 

[powerlord]

Well, it was a disaster anyway.

Firstly took till 9pm for high clouds to clear which was not forecast then

As I said - too many new things. new mount wasn't behaving and causing me lots of hassle. And then faffing around with firecapture on mac.. for some reason it kept stopping writing frames after a few seconds... i mean it's not speed - my macbook ssd can do 2GB/s write speed. No idea what the issue was. Eventually managed to get 10k frames of a section, but moon was drifting out of frame by end of 5 mins every time too, even though mount was set to lunar (is this normal ?).

So by this time I had been sitting out there in the freezing cold (literally - frost forming as I sit there) for 2 hours and had had enough.

Tried some normal imaging then, but with the C9.25 even with heater on corrector it was dewing up around the secondary. gave in and brought everything inside. sigh.

TRYING to process now but so far not having a lot of luck getting anything out of the 40GB of frames I did get 😞

Tonight looks good, but gonna stick to DSOs - much easier.