Starting with narrowband

[ste7e]

Hi all

I've recently switched from using a DSLR to using an ASI1600MM-COOL.  I'm also on the steep part of the PI learning curve.

My early attempts with narrowband are a little disappointing.  I'm struggling to separate the signal from the noise.  My first instinct is that I'm underexposing my narrowband shots so the good data is way down in the noisy regions.  Could that be where I'm going wrong?

I'm attaching a drizzle-integrated image from 12 x 600s Ha subs.  5 of the subs were at 1x1 binning and the other 7 at 2x2 binning.  The camera was running at -10C and I'm using a gain of 200.

Also attached is one of the 1x1 Ha subs.

Or, maybe this is a normal level of noise and I just have to get better at dealing with it with noise reduction?  Or, maybe this is a normal level of noise level with this number of subs and I just have to spend more time getting more shots?

As you can tell I'm floundering around a bit and not sure which way I need to go.  So, any advice at all is going to gratefully received

20170422_200_600sec_-10C_1x1_0003_Ha.fit

20170422_M51_Ha_driz.fit

[Xplode]

I haven't used this camera so i'm not familiar with the noise on it, but from looking at the 600s sub i doesn't seem unusually noisy, but the signal is very low when i compare to my own images.

Did you use any darks, flats or bias when stacking?

I can see that your stars are elongated, if you can't get round stars you need to use a lower exposure time to keep the detail in the image, cutting exposure time in 1/2 or 1/4 shouldn't be a problem with your camera. Elongated stars means the signal gets smeared out.

How often do you dither?

You should have 30+ subs and dither often for good results, with the number of images you got you're better of not using drizzle as it lowers SNR.

 

Do you use a coma corrector?

I would try to cool it more if i were you, delta T is 40-45C on these cameras so -20C should be no problem.

[ste7e]

3 minutes ago, Xplode said:

I haven't used this camera so i'm not familiar with the noise on it, but from looking at the 600s sub i doesn't seem unusually noisy, but the signal is very low when i compare to my own images.

Did you use any darks, flats or bias when stacking?

I can see that your stars are elongated, if you can't get round stars you need to use a lower exposure time to keep the detail in the image, cutting exposure time in 1/2 or 1/4 shouldn't be a problem with your camera. Elongated stars means the signal gets smeared out.

How often do you dither? Do you use a coma corrector?

I would try to cool it more if i were you, delta T is 40-45C on these cameras so -20C should be no problem.

Yep, used a master dark, bias and flat. 

My workflow is to calibrate with these using ImageCalibration first.  Then use CostmeticCorrection to try to get rid of hot and cold pixels (though I don't see a lot of improvement). Next I register the images with StarAlignment and then integrate with ImageIntegration and DrizzleIntegration.

Only just discovered the joys of dithering - since taking these shots - so I'll be doing that on the next attempt.  I use SGP which gives me the option of dithering between exposures, so that's my plan.

No, I don't have a coma corrector.

Thanks for the suggestions.  I'll have a go at halving my exposure time next time, then, but won't that knock my signal levels right down into the murky, noisy, depths?  I'll also try the camera down at -20C.

Steve

 

[Xplode]

Drizzling on images that aren't dithered is pointless, the same goes for images with elongeted stars. You really need a lot of sharp dithered images to get anything out of it, 30+ is my sduggestion.

Remember drizzling is for getting out details past the resolution of what your system can capture with good tracking, if your stars aren't sharp and round you will only get the negative effects that comes with this method.

 

When your tracking isn't accurate that lowers SNR a lot and makes the longer subs pointless, it's better with 2-3 shorter subs in the same time so you capture details, that also helps stack out noise.

You should look into taking better flats because they aren't doing what they are supposed to, the bright spots in the single image is even worse in the stacked imaged.

 

For your current data i suggest you stack it at the resolution of the 2x2 subs and no drizzle, also try stacking without the flats to see if it looks better or worse.

I bet this will give you a much better result

[ste7e]

59 minutes ago, Xplode said:

Drizzling on images that aren't dithered is pointless, the same goes for images with elongeted stars. You really need a lot of sharp dithered images to get anything out of it, 30+ is my sduggestion.

Remember drizzling is for getting out details past the resolution of what your system can capture with good tracking, if your stars aren't sharp and round you will only get the negative effects that comes with this method.

 

When your tracking isn't accurate that lowers SNR a lot and makes the longer subs pointless, it's better with 2-3 shorter subs in the same time so you capture details, that also helps stack out noise.

You should look into taking better flats because they aren't doing what they are supposed to, the bright spots in the single image is even worse in the stacked imaged.

 

For your current data i suggest you stack it at the resolution of the 2x2 subs and no drizzle, also try stacking without the flats to see if it looks better or worse.

I bet this will give you a much better result

OK, had a go with your suggestions.

The attached is the same subs stacked at the 2x2 resolution (2328 x 1760) with no flats.  You're right that it is much better.

Do you think this is as good as I'm going to get without improving my tracking.  I will try going for shorter exposures next time.

20170422_M51_Ha_lowres.fit

[MarsG76]

My 25cents worth is that when I image in NB, I always use low ISO/Gain for Ha/O3 and double gain for SII and very long exposures, 30-45 minute subs, depending on the object imaged. I tend to get good SNR so perhaps 600s is not long enough.

 

[Xplode]

That looks much better!

Improving tracking or finding an exposure time that works for your setup without getting elongated stars will help a lot, also MORE exposure time

2-3 hours of subs at 1x1 with round stars will get your very good results.

 

I have attached a 300s sub og M51 Ha from my setup, i have a faster scope, but a noisier CCD camera.

M51_Light_Ha_300sec_2017-03-23_frame5.fit

 

2 minutes ago, MarsG76 said:

My 25cents worth is that when I image in NB, I always use low ISO/Gain for Ha/O3 and double gain for SII and very long exposures, 30-45 minute subs, depending on the object imaged. I tend to get good SNR so perhaps 600s is not long enough.

 

His setup doesn't need that long subs with a fast newton and a CMOS sensor with low noise, especially not for a galaxy that has just bright blobs of Ha we're after.

CMOS sensors doesn't need as long exposures as CCD sensors becuause of the much lower read noise for CMOS.

 

 

[MarsG76]

18 minutes ago, Xplode said:

That looks much better!

Improving tracking or finding an exposure time that works for your setup without getting elongated stars will help a lot, also MORE exposure time

2-3 hours of subs at 1x1 with round stars will get your very good results.

 

I have attached a 300s sub og M51 Ha from my setup, i have a faster scope, but a noisier CCD camera.

M51_Light_Ha_300sec_2017-03-23_frame5.fit

 

His setup doesn't need that long subs with a fast newton and a CMOS sensor with low noise, especially not for a galaxy that has just bright blobs of Ha we're after.

CMOS sensors doesn't need as long exposures as CCD sensors becuause of the much lower read noise for CMOS.

 

 

I do use a CMOS sensor, but also image at 2032mm at F10 so that would increase necessary exposure time... 

[Xplode]

2 hours ago, MarsG76 said:

I do use a CMOS sensor, but also image at 2032mm at F10 so that would increase necessary exposure time... 

A f10 scope is 4x slower than a f5 + you use an old noisy camera so yeah i understand why you need so long exposures.

Even with a noisy camera and slow optics you seem to get good results with your setup 

[ste7e]

19 hours ago, Xplode said:

That looks much better!

Improving tracking or finding an exposure time that works for your setup without getting elongated stars will help a lot, also MORE exposure time

2-3 hours of subs at 1x1 with round stars will get your very good results.

 

I have attached a 300s sub og M51 Ha from my setup, i have a faster scope, but a noisier CCD camera.

M51_Light_Ha_300sec_2017-03-23_frame5.fit

 

His setup doesn't need that long subs with a fast newton and a CMOS sensor with low noise, especially not for a galaxy that has just bright blobs of Ha we're after.

CMOS sensors doesn't need as long exposures as CCD sensors becuause of the much lower read noise for CMOS.

 

 

Thanks for putting me on the right track with this, Ole.

Last night I grabbed 15 Ha 180s subs at 1x1 with dithering and I think they're looking better.  The attached is an integrated version

20170504_M51_Ha.fit

[Xplode]

Much better 

Now it's just about fine tuning tracking and getting lots of subs!

If you want to get to the next level for the quality of your subs you need a coma corrector, i can recommend the Baader MPCC III as it doesn't suffer from the reflection problems of other coma correctors.

 

I believe you can do even shorter exposures too, maybe 60s? There is an article about gain on your camera here: https://astronomy-imaging-camera.com/cooled-cameras/cooled-asi-camera-setting-in-ascom-driver/

 

[MarsG76]

On 5/5/2017 at 09:29, Xplode said:

A f10 scope is 4x slower than a f5 + you use an old noisy camera so yeah i understand why you need so long exposures.

Even with a noisy camera and slow optics you seem to get good results with your setup 

Thanks... I guess its patience and good noise reduction software.

[ste7e]

On ‎06‎/‎05‎/‎2017 at 01:27, Xplode said:

Much better 

Now it's just about fine tuning tracking and getting lots of subs!

If you want to get to the next level for the quality of your subs you need a coma corrector, i can recommend the Baader MPCC III as it doesn't suffer from the reflection problems of other coma correctors.

 

I believe you can do even shorter exposures too, maybe 60s? There is an article about gain on your camera here: https://astronomy-imaging-camera.com/cooled-cameras/cooled-asi-camera-setting-in-ascom-driver/

 

I tried a set of 60s exposures and that worked well.  Thanks for the help.

This is where I've got to now.  I'm still struggling to tame the noise but I think that's probably just down to my lack of skill with PI.  I resorted to an AdaptiveStretch to drop the noise out in the end and then a little colour correction in PS.  It looks over-processed now - I guess I just need to get more subtle...

[Xplode]

Pretty good

I don't think you should worry too much about noise, it usually looks better with a little noise than an overdone noise removal.

I see your colors aren't aligned perfectly and it lowers the detail a lot, during staralignment you might want to enable distortion correction and registration model>2d surface splines, it should align your subs better

[Adam J]

 

On 04/05/2017 at 17:30, ste7e said:

Hi all

I've recently switched from using a DSLR to using an ASI1600MM-COOL.  I'm also on the steep part of the PI learning curve.

My early attempts with narrowband are a little disappointing.  I'm struggling to separate the signal from the noise.  My first instinct is that I'm underexposing my narrowband shots so the good data is way down in the noisy regions.  Could that be where I'm going wrong?

I'm attaching a drizzle-integrated image from 12 x 600s Ha subs.  5 of the subs were at 1x1 binning and the other 7 at 2x2 binning.  The camera was running at -10C and I'm using a gain of 200.

Also attached is one of the 1x1 Ha subs.

Or, maybe this is a normal level of noise and I just have to get better at dealing with it with noise reduction?  Or, maybe this is a normal level of noise level with this number of subs and I just have to spend more time getting more shots?

As you can tell I'm floundering around a bit and not sure which way I need to go.  So, any advice at all is going to gratefully received

20170422_200_600sec_-10C_1x1_0003_Ha.fit

20170422_M51_Ha_driz.fit

Actually i was thinking that I never heard of anyone using 600 second subs on a ASI1600mmc last time i looked at the ASI1600mmc thread they were all using 60s to 120s subs and lots of them like sometimes 100 subs at 60 seconds. My impression of the camera is that is has very very low read noise but not so great on the thermal noise in longer exposures and that means that people get the best results by stacking huge numbers of shorter subs 600s seems way too long if anything.

 

Actually it appears you already worked that out.

[ste7e]

On ‎10‎/‎05‎/‎2017 at 19:17, Xplode said:

Pretty good

I don't think you should worry too much about noise, it usually looks better with a little noise than an overdone noise removal.

I see your colors aren't aligned perfectly and it lowers the detail a lot, during staralignment you might want to enable distortion correction and registration model>2d surface splines, it should align your subs better

Well spotted on the bad alignment.  I've re-aligned and re-integrated the channels now.

I've still dropped the background to black - I'd prefer a natural looking grey but I can't get a smooth texture.  The noise is just too much for my liking and NR just turns it into blobs.

Much happier with this version, though.  And, yes, I probably need to get a coma corrector at some point but I've been delaying the expensive thinking it's a lot of money for a small difference - and always had it at the back of my mind that I can probably do the correction in software in post-processing, anyway.

 

[Xplode]

A coma corrector makes more difference than you think. There is no way to get back the sharpness lost without a coma corrector by using software. Usually the middle of the field gets a little sharper too.

It also removes problems you might get by aligning images from several nights since stars won't be round and they will have different shapes unless the FOV is pretty much excactly the same.

[Allinthehead]

Hi steve. I've been using this camera since last September, and have been getting decent results with 60 second subs, (which i'm limited to as i don't guide) and high gain 399 for narrowband, and unity for broadband. The downsides to this are a huge amount of subs, 3-9 hundred, dynamic range is also reduced as you increase the gain. If i was guiding i think 2-4 minute subs would be a good compromise depending on target.

This is 6 hrs in 60 second subs. As i type i'm stacking 570 60 second subs of the Iris Nebula. It takes a couple of hours to stack but runs in the background, so not a big deal. Anyway experiment is what i'd say, and find what works for you, and your setup.

Richard.

 

[ollypenrice]

On 12/05/2017 at 13:04, ste7e said:

Well spotted on the bad alignment.  I've re-aligned and re-integrated the channels now.

I've still dropped the background to black - I'd prefer a natural looking grey but I can't get a smooth texture.  The noise is just too much for my liking and NR just turns it into blobs.

Much happier with this version, though.  And, yes, I probably need to get a coma corrector at some point but I've been delaying the expensive thinking it's a lot of money for a small difference - and always had it at the back of my mind that I can probably do the correction in software in post-processing, anyway.

 

To my eye that looks very black clipped. You've discarded a large amount of good faint data this way. I can't believe it's the way forward for you.

When you make your flats do you calibrate them? Most people simply make a master bias and use that as a dark for their flats. Because I haven't used a CMOS camera I can't be sure this will work for your setup but I think it should. Uncalibrated flats tend to over correct.

Olly

[ste7e]

On ‎20‎/‎05‎/‎2017 at 09:45, ollypenrice said:

To my eye that looks very black clipped. You've discarded a large amount of good faint data this way. I can't believe it's the way forward for you.

When you make your flats do you calibrate them? Most people simply make a master bias and use that as a dark for their flats. Because I haven't used a CMOS camera I can't be sure this will work for your setup but I think it should. Uncalibrated flats tend to over correct.

Olly

Hi Olly

Sorry it's taken me so long to respond - been away from SGL for way too long...

I do calibrate my flats.  I calibrate them with a master bias and with a master dark.

The master dark I use for all my calibration is from a set of 120 sec darks and I rely on PI to adjust those for the different exposures.  Is that a mistake?

I take it that using the bias as the dark for the flats is because of the short exposure time of the flats?

I'm trying to improve my technique for taking the flats so maybe that'll help.  I recently had a go at the Leo Triplet with better flats and think it helped - image attached.

Where I seem to be having most problems is getting a clean background in my separate mono integrated images - R, G, B and NB channels.  So, when I then combine these I get allsorts of different colours in the background where the prominence in the background for one colour doesn't match the prominence for the other colours.  I'm not sure I'm explaining this very well.  I have gradients in the background of, say the blue integrated image and a different gradient in the background of, say, the red.  So, when I combine them I get a blue cast in one area and a red cast in another.  I'm putting these gradients down to noise but maybe it's from some bad practice that's I'm polluting my workflow with?

Anyway, that Leo Triplet shot:

 

 

[ste7e]

On ‎13‎/‎05‎/‎2017 at 17:03, Xplode said:

A coma corrector makes more difference than you think. There is no way to get back the sharpness lost without a coma corrector by using software. Usually the middle of the field gets a little sharper too.

It also removes problems you might get by aligning images from several nights since stars won't be round and they will have different shapes unless the FOV is pretty much excactly the same.

Hi Ole

I have a Baader MPCC III on order that should be with me tomorrow