Sh2-129 and OU4; "Flying Bat and Squid Nebulae" OII capture question - Forums

Topic starter

Jul 20, 2022

Have been capturing some data for Sh2-129 and OU4; "Flying Bat and Squid Nebulae" in narrowband. Have about 3 hours of SII and 3 hours HA data and around 10 hours of OIII all with 300s exposures at unity. Think I underestimated how faint the OIII data is for OU4. So now I plan to capture at least another 15 hours of OIII, but the question is would it make sense to do this in 300s exposures or would I get better results by changing to 600s exposures?

Thanks for your input and thoughs

Björn Arnold

Jul 20, 2022

Hi Sean,

The primary factor is total integration time and therefore the primary goal is to do what you are planning: capture much more data.

The SNR of a 600s exposure will be better than of a 300s sub. In order to achieve a good total SNR after integration, it depends on the SNR which you have already in the present 300s exposure. If the contribution of read noise in relation to the sky background signal is already below 10% or even 5%, you could continue with the 300s.

I don’t know your equipment but often the limiting factor is the equipment. The longer the sub the better the rig has to perform (guiding etc.)

If you want to know more about SNR and single exposure time, I recommend looking at Robin Glovers (SharpCap) descriptions of how to select the exposure time. Let me know if you need some links to this.

CS, Björn

Well Written Helpful Insightful Respectful Supportive

Steven

Jul 20, 2022

Yea it's a tricky target. The Oiii data is very faint and hard to capture.
Increasing to 600 seconds might help.. but that's still a bit depending on other conditions. As the conditions and our own oxygen in our own atmosphere affect Oiii data more than Ha and Sii. - So aim for the best conditions, best visibility, darkest sky (Try to shoot Oiii in the new moon, never with a full moon), etc, etc. - obviously, you can try 600 seconds if your tracking/guiding allows for it.

I had the same camera for that target, the 183MM. It's a great camera overall, but dear god it's tricky to get light into those tiny pixels sometimes. Especially at slow apertures (guessing you're using the Z73 for this target).

Even after 20+ hours I could barely see the Oiii data with my F5 scope. I did see an improvement in shooting the Oiii data after I upgraded from a 7nm to 3nm Oiii filter. But, it's still not easy. - can't wait to try it with a faster aperture and better camera.

this one probably is the best example for you. 26+ hours with a 183MM and Z73. And the Oiii is still barely visible.
It simply is a very difficult target.

Helpful Insightful Engaging Supportive

Sean van Drogen

Topic starter

Jul 20, 2022

Yea it's a tricky target. The Oiii data is very faint and hard to capture.
Increasing to 600 seconds might help.. but that's still a bit depending on other conditions. As the conditions and our own oxygen in our own atmosphere affect Oiii data more than Ha and Sii. - So aim for the best conditions, best visibility, darkest sky (Try to shoot Oiii in the new moon, never with a full moon), etc, etc. - obviously, you can try 600 seconds if your tracking/guiding allows for it.

I had the same camera for that target, the 183MM. It's a great camera overall, but dear god it's tricky to get light into those tiny pixels sometimes. Especially at slow apertures (guessing you're using the Z73 for this target).

Even after 20+ hours I could barely see the Oiii data with my F5 scope. I did see an improvement in shooting the Oiii data after I upgraded from a 7nm to 3nm Oiii filter. But, it's still not easy. - can't wait to try it with a faster aperture and better camera.

this one probably is the best example for you. 26+ hours with a 183MM and Z73. And the Oiii is still barely visible.
It simply is a very difficult target.

Hi Steven, you guessed correctly i am using the ZS73 with the 183MM for this target. I can make out the squid in the current OIII data but have to overstretch by a whole lot to get it to pop out even a little. Tracking and guiding allows for 600sec subs. As for when I capture is mostly guided by when I get clear skies dont have lot of luxury there. Guessing this will become a multiyear project for me.

Thanks for the insight.

Sean van Drogen

Topic starter

Jul 20, 2022

Björn Arnold:
Hi Sean,

The primary factor is total integration time and therefore the primary goal is to do what you are planning: capture much more data.

The SNR of a 600s exposure will be better than of a 300s sub. In order to achieve a good total SNR after integration, it depends on the SNR which you have already in the present 300s exposure. If the contribution of read noise in relation to the sky background signal is already below 10% or even 5%, you could continue with the 300s.

I don’t know your equipment but often the limiting factor is the equipment. The longer the sub the better the rig has to perform (guiding etc.)

If you want to know more about SNR and single exposure time, I recommend looking at Robin Glovers (SharpCap) descriptions of how to select the exposure time. Let me know if you need some links to this.

CS, Björn

Thanks Björn, that makes perfect sense. Will probably stick with the 300s subs just to make processing more straight forward.

Well Written

Björn Arnold

Jul 20, 2022 · Edited: Jul 20, 2022

Sean van Drogen:
Thanks Björn, that makes perfect sense. Will probably stick with the 300s subs just to make processing more straight forward.

I checked what equipment you use. If you‘re shooting under a Bortle 8 sky like you mention in one of your images, you should be perfectly fine with the 300 second exposures.

Björn

Well Written Concise Supportive

Stefano Ciapetti

Jul 20, 2022 · Edited: Jul 20, 2022

I think sky quality is also very important and instrument size. I have done a monstre 127 hours on that subject from my bortle 8 sky, with an ASI 533, an optolong l-extreme and a Canon 70-200 F4, but nothing was visible. Subs were 60 seconds unguided:

https://astrob.in/tl6380/F/

I tried to image same subject with an 8.5 nm O3 filter, a QHY 183 mono with a 250 mm F 3.9 Newton, and something was visible, with approximately 29 hours exposure. Still bortle 8 sky.

https://astrob.in/8c0co2/0/

I think that from a city sky that object is almost impossibile. My suggestion is to try it in a very dark sky.

Helpful

Sean van Drogen

Topic starter

Jul 20, 2022

Björn Arnold:
Sean van Drogen:
Thanks Björn, that makes perfect sense. Will probably stick with the 300s subs just to make processing more straight forward.

I checked what equipment you use. If you‘re shooting under a Bortle 8 sky like you mention in one of your images, you should be perfectly fine with the 300 second exposures.

Björn

Thanks and i should have probably given more details on my specific situation but have to applaud your investigative approach.
This is the highly overstretched data so there is something there but need way more time

Tayson

Jul 20, 2022

47h oxygen ;)
https://www.astrobin.com/vut72d/C/

Sean van Drogen

Topic starter

Jul 20, 2022

Tayson:
47h oxygen ;)
https://www.astrobin.com/vut72d/C/

Thats a very impressive shot indeed. Will have my fingers crossed for many clear nights to get close that amount of data

John Noble

Jul 20, 2022 · Edited: Jul 20, 2022

Sean,

Lots of good feedback above all of which I'd agree with. I just tried imaging this target on my summer holiday from a Bortle 3 sky with a Z61/FLT 91 was underwhelmed by the O3 result even after 8 hours. That said I think the key is in how one combines the O3 with the balance of the data. The squid is so faint my assumption has been that to make it stand out you need to stretch it and denoise it separate from the rest of the channels including the stars on the Blue or O3 channels. Then blend it back in.

There is a data bundle for this target on Telescope Live and I was lucky enough to down load that a few weeks back. It contains 40+ hours of O3 shot from very dark skies and even that signal is very faint. Clearly it was better than what I had already but I was still surprised by just how faint it was.

So bottom line the O3 is super faint and my gut tells me that the best renditions are more linked to really good stretching and blending of the O3 signal - of course once you have enough data to define the squid.

Good luck

John

Helpful Insightful Respectful Engaging Supportive

Sean van Drogen

Topic starter

Jul 20, 2022

John Noble:
Sean,

Lots of good feedback above all of which I'd agree with. I just tried imaging this target on my summer holiday from a Bortle 3 sky with a Z61/FLT 91 was underwhelmed by the O3 result even after 8 hours. That said I think the key is in how one combines the O3 with the balance of the data. The squid is so faint my assumption has been that to make it stand out you need to stretch it and denoise it separate from the rest of the channels including the stars on the Blue or O3 channels. Then blend it back in.

There is a data bundle for this target on Telescope Live and I was lucky enough to down load that a few weeks back. It contains 40+ hours of O3 shot from very dark skies and even that signal is very faint. Clearly it was better than what I had already but I was still surprised by just how faint it was.

So bottom line the O3 is super faint and my gut tells me that the best renditions are more linked to really good stretching and blending of the O3 signal - of course once you have enough data to define the squid.

Good luck

John

That’s indeed a very good point so far I was using my standard SHO processing flow, but it might be good idea to use a different flow.
Thanks for the pointer

Björn Arnold

Jul 20, 2022 · Edited: Jul 20, 2022

John Noble:
So bottom line the O3 is super faint and my gut tells me that the best renditions are more linked to really good stretching and blending of the O3 signal - of course once you have enough data to define the squid.

I would only partially agree. Of course, the way you postprocess data matters a lot but a good data basis is a solid foundation. You should see the [OIII] structure clearly, even if you just have the single channel. That's independent of the what you want to blend with.
The workflow may be unusual but you might want to stretch the different channels before you combine them in order to maximise contrast in the [OIII].

The difficulty is that the actual signal you're looking for may be just a fraction compared to other signals like light poluttion which the filter still transmits + possible moonlight (e.g., under bright moon) + dark current + read noise. All these factors have a mean value and if you had no noise, you could easily postprocess it but the noise makes it difficult. Therefore, there will be a lot of data to be integrated to get the noise down to a level that one can distinguish the faint object signal from the rest.

@Sean van Drogen could you provide a calibrated fits of the [OIII] data (single sub could suffice)? Maybe based on this one can try to estimate the required integration time.

Björn

Helpful Insightful Respectful Engaging

Sean van Drogen

Topic starter

Jul 20, 2022

Björn Arnold:
John Noble:
So bottom line the O3 is super faint and my gut tells me that the best renditions are more linked to really good stretching and blending of the O3 signal - of course once you have enough data to define the squid.

I would only partially agree. Of course, the way you postprocess data matters a lot but a good data basis is a solid foundation. You should see the [OIII] structure clearly, even if you just have the single channel. That's independent of the what you want to blend with.
The workflow may be unusual but you might want to stretch the different channels before you combine them in order to maximise contrast in the [OIII].

The difficulty is that the actual signal you're looking for may be just a fraction compared to other signals like light poluttion which the filter still transmits + possible moonlight (e.g., under bright moon) + dark current + read noise. All these factors have a mean value and if you had no noise, you could easily postprocess it but the noise makes it difficult. Therefore, there will be a lot of data to be integrated to get the noise down to a level that one can distinguish the faint object signal from the rest.

@Sean van Drogen could you provide a calibrated fits of the [OIII] data (single sub could suffice)? Maybe based on this one can try to estimate the required integration time.

Björn

Hi Björn,

Here 2 calibrated and cos corrected subs. One is the best according to subframeselector, the other an average one
https://drive.google.com/drive/folders/1mm-qKUkPZ-zi0NbPZgTlW9_xPPdCFuQH?usp=sharing

Björn Arnold

Jul 20, 2022

Sean van Drogen:
Hi Björn,

Here 2 calibrated and cos corrected subs. One is the best according to subframeselector, the other an average one
https://drive.google.com/drive/folders/1mm-qKUkPZ-zi0NbPZgTlW9_xPPdCFuQH?usp=sharing

Hi Sean,

I've taken a look at the subs. From the single subs, the accuracy would be too low to estimate a proper integration time. Possibly with the integration you have already. A first estimation would match the exposure times others have mentioned with smaller refractors: 40h+

Björn

Roger Nichol

Jul 20, 2022

It is a tough target and your result will also depend on good processing techniques. I captured mine with 21 hours using an L-eXtreme filter on an OSC, the result was reasonable, but I will be going back to add another 20 or 30 hours once my current target is done. (https://www.astrobin.com/3hrk73/) (Bortle 5 location). I used 20-minute exposures for mine - that just works better for me with my setup than 4x the number of 300s exposures.

With mono narrowband, I would expect 20 hours of Oiii to be looking pretty good, although if you are in Bortle 8, then that may need more. Are you using PixInsight? If so, I would recommend using Normalise Scale Gradient to integrate the subs, carefully use Generalised Hyperbolic Stretch and then mask the Oiii and use curves to pull out the squid.

Helpful Insightful Respectful Engaging Supportive

Sean van Drogen

Topic starter

Jul 20, 2022

Roger Nichol:
It is a tough target and your result will also depend on good processing techniques. I captured mine with 21 hours using an L-eXtreme filter on an OSC, the result was reasonable, but I will be going back to add another 20 or 30 hours once my current target is done. (https://www.astrobin.com/3hrk73/) (Bortle 5 location). I used 20-minute exposures for mine - that just works better for me with my setup than 4x the number of 300s exposures.

With mono narrowband, I would expect 20 hours of Oiii to be looking pretty good, although if you are in Bortle 8, then that may need more. Are you using PixInsight? If so, I would recommend using Normalise Scale Gradient to integrate the subs, carefully use Generalised Hyperbolic Stretch and then mask the Oiii and use curves to pull out the squid.

Hi Roger,
Yes I am using PI for all my processing and use NSG script to stack. Have GHS script but have not invested the time to learn that one properly. Will put that one my to do list.

Think my first prio will now be to add a significant amount of additional data aiming for at least 25 hours OIII data. Which I my case means I will need another 4-5 clear nights.

@Björn Arnold think the message is clear need way more exposure time, fingers crossed for more clear nights

will come back once I have the data

Tim Hawkes

Jul 20, 2022 · Edited: Jul 20, 2022

This target is out of my league tough but would have thought that logically seeing the OIII nebula well must all be about SNR?

So - I guess that in theory - every Bortle unit darker the observing site –> ~ 2.5X better SNR whereas doubling the total accumulated exposure time only helps by a factor of ~ 1.4 at best. Another factor must be the width of the OIII filter window – a 3 nm one stands to be a > 2X improvement over a 6 nm version.

So a darker observation site , a new moon and a narrower OIII filter might be the most helpful (but maybe not possible?) - obviously more accumulation time helps too - but maybe it would also be possible to bin the pixels and get a further gain in SNR - since the object is quite extended?

Tim

Sean van Drogen

Topic starter

Jul 31, 2022

just a small update https://www.astrobin.com/95zh5c/what i managed so far. Need more data but weather forecasts are no good for the next few weeks.

Victor Van Puyenbroeck

Jul 31, 2022

I think your version looks very 'natural' to me and the real structure of the nebula is respected in your image.

Don't be discouraged by images you see from others, especially when you consider the limitations of your system. You're photographing from one of the most light polluted places on earth (Amsterdam) and there's no astronomical darkness around this time of the year in that latitude. Many photographers on AstroBin have much better sky conditions. I know some people also manually paint a contrast mask on top of OU4 during post-processing, to selectively brighten the whole squid-shaped area.

Some very impressive images were linked in replies above. I'd like to share less stellar results from my location near Leuven, Belgium. Hopefully that helps to put thing in perspective. Try to find the squid

7.5h OIII 3nm @f/4 and 7h Ha 3nm @f/4I found star removal very beneficial to improve visibility of the squid in the mono images.

Helpful Respectful Engaging Supportive

Sean van Drogen

Topic starter

Jul 31, 2022

Victor Van Puyenbroeck:
I think your version looks very 'natural' to me and the real structure of the nebula is respected in your image.

Don't be discouraged by images you see from others, especially when you consider the limitations of your system. You're photographing from one of the most light polluted places on earth (Amsterdam) and there's no astronomical darkness around this time of the year in that latitude. Many photographers on AstroBin have much better sky conditions. I know some people also manually paint a contrast mask on top of OU4 during post-processing, to selectively brighten the whole squid-shaped area.

Some very impressive images were linked in replies above. I'd like to share less stellar results from my location near Leuven, Belgium. Hopefully that helps to put thing in perspective. Try to find the squid 7.5h OIII 3nm @f/4 and 7h Ha 3nm @f/4I found star removal very beneficial to improve visibility of the squid in the mono images.

Hi Victor,

Thanks for the perspective. In your OIII I can clearly see the squid, very nice result.
Weather permitting I will go ahead and more data, but the biggest win is probably to be gained in the post processing.

CS Sean

Sean van Drogen

Topic starter

Aug 7, 2022

Think I am calling this one for the moment. Time to move onto another target
If someone wants to have a go here are the masters https://drive.google.com/file/d/1tYCrQl-8UZR2drDhik6gbtlFknCOSf2U/view?usp=sharingthey have only been cropped otherwise untouched

Jean-Baptiste

Aug 10, 2022

Hello @Sean van Drogen
I have seen your post lately and it happens that I tried recently this target, it is famous and it was a kind of "test" for me.
I allow to post my point of view because I think that of the previous comments could need some precision from my experience :
- yes the total integration time is definitely a crucial point
I want to add this very important point from my point of view :
- the unitary time is also key, if it is too low, you will never get a decent result, whatever you total integration time is

All that Robin Glover (from SharpCap) has posted regarding minimum unitary time is very instructive, I will try to put it with my words adapted to this target :
- your need to get the signal from the squid "out of the noise", the noise may be the sky background (with its light polution) and/or the noise from the sensor
- if the sky background is too high compared to the squid signal, you will never get a decent signal, like at dawn, at a certain point, you can not get anything because the signal is lost in the sky background
- the signal from the sky background is broadband, which means that you lower this signal compared to the squid signal with a narrowband filter
- when you do so, you MUST increase the unitary time, if you don't, you will be limited, not by the sky background but by the reading noise
- this is all the objective of this "minimum unitary time" : long enough to allow the sky background to be higher that the reading noise

When you go with narrowband filter, you may be limited by the sensor noise without enough unitary time to "get the sky background", as an example, your minimum time with an L filter (300nm) can be multiplied by 50 if you use a 6nm narrowband filter
of course, with the squid, you do this during new moon.

Personnaly, I found out that looking at the number of clipped pixels after dark subscription is good indicator of where you are (pixels at zero), unfortunately, your unitary subs is not available so I can not tell you where you are .
Let me explain why I do this : when you shoot a dark with an offset, all the pixels are non-zero.
Applying a master dark on that "dark image" will give you 50% of your pixel at zero (see that with a PIxelMath formula on the raw image such as iif ($T==0, 1, $T)
When you have almost no pixels at zero, this means that your sky background is above the gaussian noise coming from the sensor

My experience : I live in a suburban area, BOuchemaine, 10km from Angers Area (Borte 4-4.5 yellow-green)
I shoot at F3 with a 2600MM and a 4nm Oiii filter (the Baader CMOS optimised ultra narrowband) (F4 newton with Nexus x0,75 reductor)
I took 5mn exposure and this was a very good "unitary time", at F4.6, it would have been better to go to 10mn subs
With that, I can already see the squid in a unitary exposure, I compared with a F5.6 APO I bought and made a 5mn test, it was hard to see anything.

Here an autostretch of this unitary (with no calibration) :

easier to see in starless after a big noise reduction :

This night I could make some test, I got 5x5mn of Oiii and 2x5mn of Ha, it was a kind of reharsal but I could get something out of it. I was very surpised to be able to get this
it is not posted on AB but you can see the full here : https://telescopius.com/pictures/view/122571/deep_sky/Sh/2-129/diffuse-nebula/by-djibi

A screen copy (so total integration time below 1 hour with 3x5mn of RGB data for stars)

After these first test, I went for it during 3 sessions, 4,5H in Ha and 9H in Oiii, I posted it recently :
https://www.astrobin.com/ka5zp0/

I am sharing this experience with you because with your Zenistar at F6, I would recommend to go much higer than 5mn subs, with your 6nm filter, I would try 10mn with no hesitation, if you could re-post your unitary target image after dark substraction, I could have a look at it regarding the "clip pixel issue"

this is just my point of view, to help in your journey, I hope it will helps

CLear skies

Sean van Drogen

Topic starter

Aug 10, 2022

Hello @Sean van Drogen
I have seen your post lately and it happens that I tried recently this target, it is famous and it was a kind of "test" for me.
I allow to post my point of view because I think that of the previous comments could need some precision from my experience :
- yes the total integration time is definitely a crucial point
I want to add this very important point from my point of view :
- the unitary time is also key, if it is too low, you will never get a decent result, whatever you total integration time is

All that Robin Glover (from SharpCap) has posted regarding minimum unitary time is very instructive, I will try to put it with my words adapted to this target :
- your need to get the signal from the squid "out of the noise", the noise may be the sky background (with its light polution) and/or the noise from the sensor
- if the sky background is too high compared to the squid signal, you will never get a decent signal, like at dawn, at a certain point, you can not get anything because the signal is lost in the sky background
- the signal from the sky background is broadband, which means that you lower this signal compared to the squid signal with a narrowband filter
- when you do so, you MUST increase the unitary time, if you don't, you will be limited, not by the sky background but by the reading noise
- this is all the objective of this "minimum unitary time" : long enough to allow the sky background to be higher that the reading noise

When you go with narrowband filter, you may be limited by the sensor noise without enough unitary time to "get the sky background", as an example, your minimum time with an L filter (300nm) can be multiplied by 50 if you use a 6nm narrowband filter
of course, with the squid, you do this during new moon.

Personnaly, I found out that looking at the number of clipped pixels after dark subscription is good indicator of where you are (pixels at zero), unfortunately, your unitary subs is not available so I can not tell you where you are .
Let me explain why I do this : when you shoot a dark with an offset, all the pixels are non-zero.
Applying a master dark on that "dark image" will give you 50% of your pixel at zero (see that with a PIxelMath formula on the raw image such as iif ($T==0, 1, $T)
When you have almost no pixels at zero, this means that your sky background is above the gaussian noise coming from the sensor

My experience : I live in a suburban area, BOuchemaine, 10km from Angers Area (Borte 4-4.5 yellow-green)
I shoot at F3 with a 2600MM and a 4nm Oiii filter (the Baader CMOS optimised ultra narrowband) (F4 newton with Nexus x0,75 reductor)
I took 5mn exposure and this was a very good "unitary time", at F4.6, it would have been better to go to 10mn subs
With that, I can already see the squid in a unitary exposure, I compared with a F5.6 APO I bought and made a 5mn test, it was hard to see anything.

Here an autostretch of this unitary (with no calibration) :

easier to see in starless after a big noise reduction :

This night I could make some test, I got 5x5mn of Oiii and 2x5mn of Ha, it was a kind of reharsal but I could get something out of it. I was very surpised to be able to get this
it is not posted on AB but you can see the full here : https://telescopius.com/pictures/view/122571/deep_sky/Sh/2-129/diffuse-nebula/by-djibi

A screen copy (so total integration time below 1 hour with 3x5mn of RGB data for stars)

After these first test, I went for it during 3 sessions, 4,5H in Ha and 9H in Oiii, I posted it recently :
https://www.astrobin.com/ka5zp0/

I am sharing this experience with you because with your Zenistar at F6, I would recommend to go much higer than 5mn subs, with your 6nm filter, I would try 10mn with no hesitation, if you could re-post your unitary target image after dark substraction, I could have a look at it regarding the "clip pixel issue"

this is just my point of view, to help in your journey, I hope it will helps

CLear skies

Well I am in luck as some more clear nights are predicted so I will follow your suggestion and see what 10 minute subs can add to this. I shoot from within Amsterdam right next to the Football Arena, but if i pick targets in the opposite direction and high enough its luckily not too much trouble.

Thanks for the very interesting and detailed answer

Dale A Chamberlain

Aug 26, 2022

Sharpless 129 and Ou4 "Flying Bat" and "Squid" nebulae
I just did this under Bortle 5 and nearly moonless skies. Ha was 3 hours and OIII was 9. And my subs were 90 seconds each.