Astrobiscuit is looking for quality astrophotographers to join the Big Amateur Telescope (BAT)

[rorymultistorey]

On 02/07/2021 at 12:09, vlaiv said:

Are you sure about this?

Back of the envelope calculation suggests that you'll be getting about 0.4e/px/s from sky. In 5s exposure that means about 2e of background signal or about 1.41e of LP noise. Hardly swamping 1e of read noise.

You need about 1 minute of exposure to truly swamp read noise (5:1 - LP to read).

 

I just like to say to other readers unsure about whether lucky imaging works or not to check galaxies on astrobin and see how the sharpest ones had exposure lengths of just a few seconds.  The proof is in the pudding.

 

 

Edited July 4, 2021 by rorymultistorey
Astrobiscuit (me) said something stupid and offensive so I've removed it and apologise.

[david_taurus83]

2 hours ago, rorymultistorey said:

I just like to say to other readers unsure about whether lucky imaging works or not to check galaxies on astrobin and see how the sharpest ones had exposure lengths of just a few seconds.  The proof is in the pudding.

Vlaiv seem to be very keen to pick holes in everything I or anyone else says. I don't claim to know everything but i'm not sure you know everything either.  If I had to guess I'd say you were an armchair astronomer who hasn't actually been out their and taken photos. I could be wrong but that is what I'm thinking. In the case of the read noise have you considered  the beautiful and powerful noise reduction that comes from stacking thousands of subs. Whatever your answer I will simply come back and say that I've done it and it works. And I did it with a £700 set up.  Lucky imaging is not easy, lucky imaging relies on lots of other things being set up correctly, lucky imaging is continually fighting the fact that we're not catching many photons but it works. Its the theory of proof and pudding. So please if someone else is keen to join the BAT please don't be put off by negativity. We're a lovely bunch and if your set up isn't quite up to lucky imaging standard then we will ask you to help us get the dim bits of whatever we're shooting bc the faint bits is where lucky imaging really struggles.

 

Hi Rory. I think your being a bit harsh on vlaiv. This is an open forum and he's well entitled to contribute. He hasn't said the project can't be done as far as I can see and he has highlighted some obstacles to overcome. 

Ps. Looking forward to the next videos and to see how this project is coming along.

[admin]

1 hour ago, david_taurus83 said:

Hi Rory. I think your being a bit harsh on vlaiv. This is an open forum and he's well entitled to contribute. He hasn't said the project can't be done as far as I can see and he has highlighted some obstacles to overcome. 

We agree. 

Anyone who has spent any time at SGL will know vlaiv is an experienced astronomer respected for his knowledge and contribution to the community. 

@rorymultistorey You are welcome to use SGL to publicise your project and raise support but we will not allow personal attacks. 

[rorymultistorey]

16 minutes ago, admin said:

We agree. 

Anyone who has spent any time at SGL will know vlaiv is an experienced astronomer respected for his knowledge and contribution to the community. 

@rorymultistorey You are welcome to use SGL to publicise your project and raise support but we will not allow personal attacks. 

I am truly sorry. Sorry vlaiv. I saw red. I suspect it has as much to do with other things happening right now rather than you. Either way I was out of order.  

[vlaiv]

15 minutes ago, rorymultistorey said:

I am truly sorry. Sorry vlaiv. I saw red. I suspect it has as much to do with other things happening right now rather than you. Either way I was out of order.  

Don't worry about that. I'm sorry if any of my comments sounded malicious in any way.

Indeed, I was trying to point out some of the things people involved in the project need to be aware of in order for project to work to its full potential.

5 hours ago, rorymultistorey said:

I just like to say to other readers unsure about whether lucky imaging works or not to check galaxies on astrobin and see how the sharpest ones had exposure lengths of just a few seconds.  The proof is in the pudding.

Vlaiv seem to be very keen to pick holes in everything I or anyone else says. I don't claim to know everything but i'm not sure you know everything either.  If I had to guess I'd say you were an armchair astronomer who hasn't actually been out their and taken photos. I could be wrong but that is what I'm thinking. In the case of the read noise have you considered  the beautiful and powerful noise reduction that comes from stacking thousands of subs. Whatever your answer I will simply come back and say that I've done it and it works. And I did it with a £700 set up.  Lucky imaging is not easy, lucky imaging relies on lots of other things being set up correctly, lucky imaging is continually fighting the fact that we're not catching many photons but it works. Its the theory of proof and pudding. So please if someone else is keen to join the BAT please don't be put off by negativity. We're a lovely bunch and if your set up isn't quite up to lucky imaging standard then we will ask you to help us get the dim bits of whatever we're shooting bc the faint bits is where lucky imaging really struggles.

I know lucky DSO imaging works. There is no question about it really. However, I'd like to point out that results that I've seen come from rather large telescopes.

Take for example Emil's work:

https://www.astrokraai.nl/viewimages.php?id=266&cd=7

There is simple explanation for this. On one hand - achieved resolution depends on aperture size. I'm not talking here about planetary critical sampling which is probably x3-4 higher than deep sky sampling. Even with DSO imaging - aperture is important for resolution. This is because different PSFs convolve image with compounding effects. One PSF is mount tracking / guiding performance. If you use short exposures, you'll be minimizing impact of that as there is less chance for mount to drift in short time. Other is seeing - that is addressed with selection of subs and choosing those where seeing impact is the least.

Third is aperture. Here - you don't really have any options except to change the scope. It's safe to assume that people won't be doing that for purpose of this project, so they are stuck with PSF of scope that they have. 4" will have double blur of 8" scope due to aperture alone, which in turn will have half that of 16" scope.

Better the seeing, or selecting best of the subs - more impact above will have on final resolution. For telescope sizes that we usually use in imaging - aperture has rather small effect compared to tracking and seeing, but once you minimize those two like in lucky imaging - then aperture suddenly becomes important.

Second important point with large aperture telescopes is - once you set your working resolution / sampling rate, speed of your system depends solely on aperture size. This enables you to get large signal in short amount of time - beneficial both for accurate stacking and also to overcome read noise.

In the end, here is interesting resource that people should examine:

https://www.meteoblue.com/en/weather/outdoorsports/seeing/london_united-kingdom_2643743

This website gives forecast of seeing in arc seconds FWHM.

It would be good to check forecast value against actual measured values. Care must be taken to subtract part that is due to aperture size. If forecast is reliable (and I think it is - at least from planetary imaging point of view and I don't see why it will be different here) - then it will be good indicator of results that can be expected.

You can't really expect to get 2" FWHM if forecast shows something like this:

and you are using 4" scope (which itself has ~1.12" FWHM contribution due to aperture).

By the way, aperture FWHM can be calculated by using 2.355 * 0.42 * lambda / aperture expression (which will give you radians if you put same units for aperture and wavelength  - you can put 550nm for wavelength), and you can combine two by adding them in quadrature:

sqrt(FWHM_seeing^2 + FWHM_aperture^2)

Here we omitting contribution due to mount and assume perfect aperture.

[Sunshine]

This sounds like a whole lot of fun, I will be following this for sure, cant wait to see the results!.

[rorymultistorey]

13 hours ago, vlaiv said:

Don't worry about that. I'm sorry if any of my comments sounded malicious in any way.

Indeed, I was trying to point out some of the things people involved in the project need to be aware of in order for project to work to its full potential.

I know lucky DSO imaging works. There is no question about it really. However, I'd like to point out that results that I've seen come from rather large telescopes.

Take for example Emil's work:

https://www.astrokraai.nl/viewimages.php?id=266&cd=7

There is simple explanation for this. On one hand - achieved resolution depends on aperture size. I'm not talking here about planetary critical sampling which is probably x3-4 higher than deep sky sampling. Even with DSO imaging - aperture is important for resolution. This is because different PSFs convolve image with compounding effects. One PSF is mount tracking / guiding performance. If you use short exposures, you'll be minimizing impact of that as there is less chance for mount to drift in short time. Other is seeing - that is addressed with selection of subs and choosing those where seeing impact is the least.

Third is aperture. Here - you don't really have any options except to change the scope. It's safe to assume that people won't be doing that for purpose of this project, so they are stuck with PSF of scope that they have. 4" will have double blur of 8" scope due to aperture alone, which in turn will have half that of 16" scope.

Better the seeing, or selecting best of the subs - more impact above will have on final resolution. For telescope sizes that we usually use in imaging - aperture has rather small effect compared to tracking and seeing, but once you minimize those two like in lucky imaging - then aperture suddenly becomes important.

Second important point with large aperture telescopes is - once you set your working resolution / sampling rate, speed of your system depends solely on aperture size. This enables you to get large signal in short amount of time - beneficial both for accurate stacking and also to overcome read noise.

In the end, here is interesting resource that people should examine:

https://www.meteoblue.com/en/weather/outdoorsports/seeing/london_united-kingdom_2643743

This website gives forecast of seeing in arc seconds FWHM.

It would be good to check forecast value against actual measured values. Care must be taken to subtract part that is due to aperture size. If forecast is reliable (and I think it is - at least from planetary imaging point of view and I don't see why it will be different here) - then it will be good indicator of results that can be expected.

You can't really expect to get 2" FWHM if forecast shows something like this:

and you are using 4" scope (which itself has ~1.12" FWHM contribution due to aperture).

By the way, aperture FWHM can be calculated by using 2.355 * 0.42 * lambda / aperture expression (which will give you radians if you put same units for aperture and wavelength  - you can put 550nm for wavelength), and you can combine two by adding them in quadrature:

sqrt(FWHM_seeing^2 + FWHM_aperture^2)

Here we omitting contribution due to mount and assume perfect aperture.

vlaiv first off thank you for being such a gentleman. Sometimes I am an idiot. I would put a ruder word in here in  place of idiot but you know kids and all...

The other thing is that it feels like you think I don't  know this stuff but I guess you're not actually talking to me. However I do worry that your missing the big point. The big point is that the biggeest source of blurryness dominates all others (thats the  sqrt(FWHM_seeing^2 + FWHM_aperture^2) equation talking)   If you boil it all down to something really simple then you can say... like I did in the video -  that the biggest source of blurryness for a telescope larger than 6 inches in diameter is -assuming its a good scope that has been well collimated -  more often than not the seeing. And as you know I'm interesting in overcoming the seeing. One of my members is directly imaging exoplanets at the keck observatory and pointed me towards one of his colleagues rather interesting scientific paper on lucky imaging. https://www.ast.cam.ac.uk/sites/default/files/nlaw_lucky_thesis.pdf We can't shoot millisecond exposures of course but even second long exposures reduce dramatically the blurryness from the seeing. 

For me its like us amateurs have only been imaging in 2nd gear. Its exciting to think that we can go much further. Many of our members don't have scopes that are sharp enough to lucky image with (collimation is a big problem actually) and many don't have cameras with a low enough read noise to lucky image with. I'm still keen that they join in bc we will help them improve and advise them on buying better equipment and we can use their long exposure non lucky imaging data to help reveal detail in fainter structures that are too dim to see through lucky imaging. Some have the equipment but don't yet have the skills to pull off lucky imaging, again we can help them.  

This is exciting. And if us amateurs pull together and share our data we can to an extent negate the inherent problems associated with short exposure deep space astrophotography and that will give us the power - especially on the bright targets - to match or even better the 1.1arcsecond FWHM that giant scopes like the 4m Mayall on Kitt Peak achieve. 

 

Edited July 3, 2021 by rorymultistorey

[vlaiv]

27 minutes ago, rorymultistorey said:

We can't shoot millisecond exposures of course but even second long exposures reduce dramatically the blurryness from the seeing. 

Here is what I believe is wrong.

Millisecond exposures help freeze the seeing. Atmosphere is turbulent and constantly changing. Depending on aperture size - there is small window in which it is for all intents and purposes "stationary". For amateur sized telescopes in better/good seeing it is about 5-6ms (maybe even up to 10ms on nights of very good/excellent seeing).

Any longer than this and things start to average out - "motion blur" starts to happen on PSF. It is not coincidence that measure used to describe the seeing is "FWHM of two second integration". Two seconds is enough for things to average out (those millisecond changes) and provide steady seeing figure that won't change much between successive exposures.

While DSO lucky imaging works - it is still limited by this averaging. In planetary imaging (if done properly and exposure length kept short) we don't have this average - in fact, we keep only very small subset of valid frames - and even with them - we can actually choose to stack part of frame. This is because seeing PSF varies strongly over even short distances (I think about 10 arc seconds away you can have entirely different PSF - one blurred enough not to be included in stack - while other perfectly "calm"). In 2 second exposure - we can expect that FWHM of all stars in the image will have very close value because all of them have been averaged out versus short exposures.

I think this would be interesting place to put some real data in this discussion. I'm currently veryu busy - I'm finishing new house (finishing is rather strange word - better expression would be finishing just enough so I can move in ) and am about to move in next 2-3 weeks so can't be of much of help there - but I suspect you already have needed data?

How about taking one of your lucky sessions that has 1-2 seconds long exposures and doing stats on FWHM? Mean value, standard deviation of measured FWHMs - maybe even a graph of how it changes over time?

That would give people a good idea of what the can expect from lucky imaging, how much of frames can they expect to keep and to get the sense of FWHM values in general.

 

[bottletopburly]

This is going to be a really exciting project to watch unfold time to seperate the wheat from the chaff ,this  potentiallly is citizen science at its best good luck  to the team participating .

[rorymultistorey]

1 hour ago, vlaiv said:

Here is what I believe is wrong.

Millisecond exposures help freeze the seeing. Atmosphere is turbulent and constantly changing. Depending on aperture size - there is small window in which it is for all intents and purposes "stationary". For amateur sized telescopes in better/good seeing it is about 5-6ms (maybe even up to 10ms on nights of very good/excellent seeing).

Any longer than this and things start to average out - "motion blur" starts to happen on PSF. It is not coincidence that measure used to describe the seeing is "FWHM of two second integration". Two seconds is enough for things to average out (those millisecond changes) and provide steady seeing figure that won't change much between successive exposures.

While DSO lucky imaging works - it is still limited by this averaging. In planetary imaging (if done properly and exposure length kept short) we don't have this average - in fact, we keep only very small subset of valid frames - and even with them - we can actually choose to stack part of frame. This is because seeing PSF varies strongly over even short distances (I think about 10 arc seconds away you can have entirely different PSF - one blurred enough not to be included in stack - while other perfectly "calm"). In 2 second exposure - we can expect that FWHM of all stars in the image will have very close value because all of them have been averaged out versus short exposures.

I think this would be interesting place to put some real data in this discussion. I'm currently veryu busy - I'm finishing new house (finishing is rather strange word - better expression would be finishing just enough so I can move in ) and am about to move in next 2-3 weeks so can't be of much of help there - but I suspect you already have needed data?

How about taking one of your lucky sessions that has 1-2 seconds long exposures and doing stats on FWHM? Mean value, standard deviation of measured FWHMs - maybe even a graph of how it changes over time?

That would give people a good idea of what the can expect from lucky imaging, how much of frames can they expect to keep and to get the sense of FWHM values in general.

 

Yes of course.😊 The WHOLE point of the Big Amateur Telescope is to test stuff out and that is what we are setting up to do. Its a journey we are going on together. So far we have 60 amateurs so far from all over the world. Many very knowledgeable and a few with surprisingly big scopes (like 50cm).  (We might even be getting one of these big scopes paired up a large pixel low read noise full frame camera which potentially could have a go a 10ms imaging on the brighter targets- wow)

BTW according to they guy who images at keck there is still a benefit from 2 second exposures. Not as much benefit as millisecond exposures but it is significant. 

 

 

 

 

Edited July 3, 2021 by rorymultistorey
removed my image of m106

[DaveS]

Assuming it's M106 you're after, this is what Astronomy Tools says for my QHY268M on the ODK12

Totally bonkers but might be worth a laugh or two. However, at 52MB per sub it's going to add up quickly.

[DaveS]

Dynamic Range for the QHY268M

And Readout Noise

At a Gain of 60 things look *very* interesting.

Damn, I'm talking myself into doing something *Very Silly* #MontyPython

[cfinn]

50 minutes ago, DaveS said:

Dynamic Range for the QHY268M

And Readout Noise

At a Gain of 60 things look *very* interesting.

Damn, I'm talking myself into doing something *Very Silly* #MontyPython

I have read the transition to high conversion gain is actually at gain 56 (graph is a bit misleading) but I think for short exposures of a few seconds you may as well take it all the way out to gain 100 in mode 1, then you’ll be close to 1e read noise. I think the most important thing for this kind of imaging is that we minimise read noise as much as possible. Supposedly this sensor can even go down to 0.7e read noise or thereabouts at even higher gain. The Altair variant (26M) has something called “ultra high conversion gain mode” which is supposed to enable this, but I see no such option on the QHY or ZWO cameras at the moment.

[rorymultistorey]

15 minutes ago, cfinn said:

I have read the transition to high conversion gain is actually at gain 56 (graph is a bit misleading) but I think for short exposures of a few seconds you may as well take it all the way out to gain 100 in mode 1, then you’ll be close to 1e read noise. I think the most important thing for this kind of imaging is that we minimise read noise as much as possible. Supposedly this sensor can even go down to 0.7e read noise or thereabouts at even higher gain. The Altair variant (26M) has something called “ultra high conversion gain mode” which is supposed to enable this, but I see no such option on the QHY or ZWO cameras at the moment.

 

this is very interesting I didn't know about the Altair model... Also noob question what are the different modes/different coloured lines about.

Edited July 3, 2021 by rorymultistorey

[cfinn]

3 minutes ago, rorymultistorey said:

 

this is very interesting I didn't know about the Altair model... Also noob question what are the different modes/different coloured lines about.

This is it. They say read noise is “0.79e in HCG + Ultra Mode” in the text description. QHY have different read out modes for their cameras, which is what all those curves are. ZWO make it simpler with just one readout mode. For the 268M, read out mode 1 is what most people use in most circumstances, because it provides the lowest read noise. There is also this nice feature in mode 1 at gain 56, where there is a transition from low to high conversion gain. At that point exactly, the read noise drops significantly and the dynamic range is maximised. So mode 1 gain 56 is what most imagers with the 268M will use, but for lucky imaging I think we may as well go all the way to gain 100 in mode 1 to reduce read noise as much as possible. If Altair can unlock even lower read noise from this sensor then I don’t see why QHY and ZWO can’t. Fingers crossed they release it in a firmware update at some point down the line.

[vlaiv]

5 minutes ago, cfinn said:

So mode 1 gain 56 is what most imagers with the 268M will use, but for lucky imaging I think we may as well go all the way to gain 100 in mode 1 to reduce read noise as much as possible. If Altair can unlock even lower read noise from this sensor then I don’t see why QHY and ZWO can’t. Fingers crossed they release it in a firmware update at some point down the line.

I'd take published read noise results with a grain of salt.

It is best to measure it yourself rather than rely on published data. It is rather simple procedure - take two bias subs, use e/ADU factor to convert pixel values to electrons instead of ADU, subtract two subs, measure standard deviation and then divide that value with square root of two (subtracting two subs has the same effect as adding them, so sum of two subs will have their noise added - but when subtracting, you'll cancel out bias signal).

[cfinn]

1 minute ago, vlaiv said:

I'd take published read noise results with a grain of salt.

It is best to measure it yourself rather than rely on published data. It is rather simple procedure - take two bias subs, use e/ADU factor to convert pixel values to electrons instead of ADU, subtract two subs, measure standard deviation and then divide that value with square root of two (subtracting two subs has the same effect as adding them, so sum of two subs will have their noise added - but when subtracting, you'll cancel out bias signal).

Thanks vlaiv, yes, I intend to do this at some point. I saw a couple of people on cloudy nights do this very test already and confirmed the QHY published results, so I have been lazy thus far and taken their word for it.

[vlaiv]

4 minutes ago, cfinn said:

Thanks vlaiv, yes, I intend to do this at some point. I saw a couple of people on cloudy nights do this very test already and confirmed the QHY published results, so I have been lazy thus far and taken their word for it.

I've done that test with couple of ZWO cameras (different models) and it seemed that published figures are always a bit optimistic. Not by much - for example, ASI1600 that has 1.7e at unity gain was measured to be closer to 1.8e and so on.

[tomato]

Probably a question for the BAT forum, but will there be a nominated single target to image? Presumably that will be required if we are going to get sufficient data on one target, but that is going to be tricky if the imagers signed up are all over the globe.

And M106 will be poorly placed from the UK when Astro darkness returns.

[rorymultistorey]

15 minutes ago, tomato said:

Probably a question for the BAT forum, but will there be a nominated single target to image? Presumably that will be required if we are going to get sufficient data on one target, but that is going to be tricky if the imagers signed up are all over the globe.

And M106 will be poorly placed from the UK when Astro darkness returns.

Actually we're having a town hall about choosing the next target tomorrow 8pm in "the bat chat" channel. You'll need to have passed the bat exam first before you can join - which is as easy as dropping a pic that you've taken in the bat exam channel. One of our team will then let you in... hold on a minute, your already in aren't you. You should have had a notification about the meeting.

 

Edited July 3, 2021 by rorymultistorey

[DaveS]

3 hours ago, cfinn said:

I have read the transition to high conversion gain is actually at gain 56 (graph is a bit misleading) but I think for short exposures of a few seconds you may as well take it all the way out to gain 100 in mode 1, then you’ll be close to 1e read noise. I think the most important thing for this kind of imaging is that we minimise read noise as much as possible. Supposedly this sensor can even go down to 0.7e read noise or thereabouts at even higher gain. The Altair variant (26M) has something called “ultra high conversion gain mode” which is supposed to enable this, but I see no such option on the QHY or ZWO cameras at the moment.

Yep, I had been fixated on maximising DR. Maybe pushing the gain to 100 would be a good idea.

[DaveS]

Uploaded to the BAT exam.

[DaveS]

On 03/07/2021 at 13:52, cfinn said:

I have read the transition to high conversion gain is actually at gain 56 (graph is a bit misleading) but I think for short exposures of a few seconds you may as well take it all the way out to gain 100 in mode 1, then you’ll be close to 1e read noise. I think the most important thing for this kind of imaging is that we minimise read noise as much as possible. Supposedly this sensor can even go down to 0.7e read noise or thereabouts at even higher gain. The Altair variant (26M) has something called “ultra high conversion gain mode” which is supposed to enable this, but I see no such option on the QHY or ZWO cameras at the moment.

Looking at Gain=100, everything drops disastrously, DR drops below 9 stops (Looks about 8.5) and Full Well too

I think keeping Gain near 56-60 is the best bet.