So much Noise!!!

[astrosathya]

Hi Everyone,

I tried my 533MC PRO last night on M12. The camera was at 0 degree C when shooting. Below is a Debayered, ABE in Pixinsight screenshot and a closeup of a colour calibrated section. What am I doing wrong?

 I will be massively grateful for advice.

Kind regards,

[Elp]

Cooled cameras still have noise, but it's finer grain compared to uncooled cameras. It may also be a case when you were testing there was high cloud which adds to the murky look.

Pretty much all AP images have noise reduction applied post process. Spending more hours on an image helps mitigate noise overall.

Edited August 21 by Elp

[TiffsAndAstro]

Stars look elongated in the direction top left to bottom right across the entire frame. Dunno what caused that.

In the close up some of the stars look almost donuty which makes me think focus isn't quite right?

Also in first image there is most of an offset donut shape so collimation too?

Like elp said thin high cloud our eyes struggle to see make images very noisy.

on the upside, it all looks pretty close, so hopefully just few small tweaks?

Edited August 21 by TiffsAndAstro

[ONIKKINEN]

How long is the integration here? Signal to noise ratio increases as you integrate longer datasets, its hard to say without knowing how long this took. Local conditions also play a huge role in this, if you have a lot of light pollution then images are more noisy, likewise if there was poor transparency due to high cloud or fog or something. The second image looks very zoomed in, maybe even past 100% in which case every image taken with any kit under any skies is likely to look noisy.

One thing that looks to be the case from the sample images is that you were out of focus, which is why you have donut shaped hollow stars. Its not exactly related to your noise problem, but it is something you should look into.

Another thing is the weird blue ring with a dark center, although hard to tell from the screenshot. Either ABE went wrong or you dont have flats (or dont have good flats), or both. I would recommend using DBE instead of ABE, as ABE often makes mistakes and creates these sort of funny bright rings and dark spots especially if flats were not so good.

[Fegato]

Need a bit more info really.

Firstly, as mentioned, your focus looks poor. Collimation may or may not be an issue - is it good?

Were you guiding? If so, was it good? It is possible to get elongation across the frame due to poor guiding in one axis (normally RA).

And how many frames in the stack, how long were they?

BTW - I would not recommend ABE in Pixinsight, although might be fine with this image. DBE gives you much more control, and the new Gradient Correction seems to be working well for me. Doubt this causes your big blue swirl that doesn't look quite right - reflection or light leak maybe?

 

 

 

 

[astrosathya]

10 hours ago, Elp said:

Cooled cameras still have noise, but it's finer grain compared to uncooled cameras. It may also be a case when you were testing there was high cloud which adds to the murky look.

Pretty much all AP images have noise reduction applied post process. Spending more hours on an image helps mitigate noise overall.

Thank you Elp. Yes, I am aware of noise reduction processes applied during the image processing. The reason I was taken aback was because I compared this to my QHY9M CCD from 2 years ago and it took me completely by surprise. I typically spend 4-6 hours on any object, this was but a quick test of the new camera.

[astrosathya]

10 hours ago, TiffsAndAstro said:

Stars look elongated in the direction top left to bottom right across the entire frame. Dunno what caused that.

In the close up some of the stars look almost donuty which makes me think focus isn't quite right?

Also in first image there is most of an offset donut shape so collimation too?

Like elp said thin high cloud our eyes struggle to see make images very noisy.

on the upside, it all looks pretty close, so hopefully just few small tweaks?

Yes, All of the above are correct. I did not mention that collimation was slightly off (hence the offset vignette, and focus was slightly off too. Integration was merely 30s x 10 exposures, guiding was absolutely rubbish. The guide star was disco dancing all over the FoV.

I am usually not used to this much noise and that baffled me. My QHY9M images were smooth as butter. 

[astrosathya]

1 hour ago, Fegato said:

Need a bit more info really.

Firstly, as mentioned, your focus looks poor. Collimation may or may not be an issue - is it good?

Were you guiding? If so, was it good? It is possible to get elongation across the frame due to poor guiding in one axis (normally RA).

And how many frames in the stack, how long were they?

BTW - I would not recommend ABE in Pixinsight, although might be fine with this image. DBE gives you much more control, and the new Gradient Correction seems to be working well for me. Doubt this causes your big blue swirl that doesn't look quite right - reflection or light leak maybe?

 

 

 

 

Focus & Collimation were poor and guiding was horrendous to say the least. 30s x 10 exposures. I used ABE only as a quick check, I always used DBE until 2 years ago when I took a break, sold all my old kit and moved to UK  This new camera is driving me slightly nuts as of today because I was never used to this much noise. I had a QHY9M and a cooled DSLR.

[Fegato]

300s versus 6 hours will be very noisy in comparison. You need to compare similar integration times?

[astrosathya]

6 hours ago, ONIKKINEN said:

How long is the integration here? Signal to noise ratio increases as you integrate longer datasets, its hard to say without knowing how long this took. Local conditions also play a huge role in this, if you have a lot of light pollution then images are more noisy, likewise if there was poor transparency due to high cloud or fog or something. The second image looks very zoomed in, maybe even past 100% in which case every image taken with any kit under any skies is likely to look noisy.

One thing that looks to be the case from the sample images is that you were out of focus, which is why you have donut shaped hollow stars. Its not exactly related to your noise problem, but it is something you should look into.

Another thing is the weird blue ring with a dark center, although hard to tell from the screenshot. Either ABE went wrong or you dont have flats (or dont have good flats), or both. I would recommend using DBE instead of ABE, as ABE often makes mistakes and creates these sort of funny bright rings and dark spots especially if flats were not so good.

Hi, thank you very much Onikkinen. Yes, I was a bit lazy, did a quick focus, bad guiding, light leaks from the bottom of the OTA. and no flats  
I had to mention in the original post that members please ignore the rest. But I am grateful for all the replies.

[astrosathya]

1 minute ago, Fegato said:

300s versus 6 hours will be very noisy in comparison. You need to compare similar integration times?

Ofcourse, I was being lazy and only wanted to test the camera's urban legend that its completely noise free and no need to take darks. I will do 300s subs tonight and check how that goes.

[Elp]

You need to compare like for like total integration times as well as similar nights imaging conditions, you can't do a short sub vs hours comparison.

The "marketing hype" of the 533 is it's zero amp glow tech, my tests taking ten minute darks were clean in comparison to other sensors which have amp glow.

Also you should see significant differences when binning the two types of data.

 

Edited August 22 by Elp

[wimvb]

Do you have the same light pollution now as you had with the QHY? You wrote that you moved to the UK. From where? In the UK you can expect a lot of light pollution. ABE and DBE will remove not only gradients, but also any light pollution, but not the associated noise. If you lose one magnitude in sky brightness due to increased light pollution, you have to increase the total integration time by a factor of 2.5 in order to get a similar result in terms of SNR. For each hour of integration time under Mag 19 skies, you’ll need 2.5 hours under Mag 18 skies. That’s why having access to dark sites is so important.

[900SL]

3 hours ago, astrosathya said:

Ofcourse, I was being lazy and only wanted to test the camera's urban legend that its completely noise free and no need to take darks. I will do 300s subs tonight and check how that goes.

 

The noise in your image is because you only exposed for 300s. It's shot noise and read noise which appears as colour mottling. These are reduced by cumulative exposure, stacking, reduced light pollution etc. (increased signal)

It's not an urban legend that the 533 and 1600 sensors are effectively noise free. This however refers to dark current noise (amp glow).

There's a bunch of other issues in your stack, focus, calibration, collimation, coma? etc.

I suggest you read up on each subject so you understand what you are trying to do, before dismissing the camera. Decent astro images are a long haul, lot's to understand and get right

Here is a single 360s sub, and a stack of 24 x 360s subs, to illustrate. From a 533MM Pro. The stack still demonstrates noise, which is shot noise, however the signal is much higher and the noise reduced

 

 

 

Edited August 22 by 900SL

[Stuart1971]

3 hours ago, astrosathya said:

Thank you Elp. Yes, I am aware of noise reduction processes applied during the image processing. The reason I was taken aback was because I compared this to my QHY9M CCD from 2 years ago and it took me completely by surprise. I typically spend 4-6 hours on any object, this was but a quick test of the new camera.

CMOS sensors are very different from the older CCD, and what you are seeing is just not enough data, and so the noise is swamping the signal. With the same amount of data with a CCD is may well have been much cleaner, but there would not have been as much of the object showing, so with CMOS you need much more data but you can get away with shorter exposures, with CCD you really needed longer individual subs, to get more data

So very simply  put CMOS you can take loads of short exposures, and CCD the same amount of integration time but consisting of longer subs, ( to swamp the red noise)and obvioulsy with more control over the gain with CMOS cameras helps too. And also with shorter subs you can afford to lose a few if you have guiding or other issues, whereas if taking 30 min exposures with CCD you would not really want to lose any..

[Xilman]

14 hours ago, astrosathya said:

Hi Everyone,

I tried my 533MC PRO last night on M12. The camera was at 0 degree C when shooting. Below is a Debayered, ABE in Pixinsight screenshot and a closeup of a colour calibrated section. What am I doing wrong?

 I will be massively grateful for advice.

Kind regards,

Focus is way out.  You can see the shadow of the secondary in the disk image of the primary. With tight focus the donuts collapse to small bright disks.

The camera orientation is not given so we can't tell whether the elongation of the star images arises for diurnal drift (i.e. poor guiding) or astigmatism which is a common sign of poor collimation and/or focus. My telescope shows \ astigmatism one side of focus and / on the other. It's how I used to focus until a Bahtinov mask made the process so much simpler and more accurate.

The only thing you can practically do with the noise is to take longer exposures, probably by stacking more subs, or use noise-removal software in the post-processing phase. I have found FABADA to be especially effective. https://github.com/PabloMSanAla/fabada

 

 

[astrosathya]

1 hour ago, 900SL said:

 

The noise in your image is because you only exposed for 300s. It's shot noise and read noise which appears as colour mottling. These are reduced by cumulative exposure, stacking, reduced light pollution etc. (increased signal)

It's not an urban legend that the 533 and 1600 sensors are effectively noise free. This however refers to dark current noise (amp glow).

There's a bunch of other issues in your stack, focus, calibration, collimation, coma? etc.

I suggest you read up on each subject so you understand what you are trying to do, before dismissing the camera. Decent astro images are a long haul, lot's to understand and get right

Here is a single 360s sub, and a stack of 24 x 360s subs, to illustrate. From a 533MM Pro. The stack still demonstrates noise, which is shot noise, however the signal is much higher and the noise reduced

 

 

Hi, thanks for the indepth explanation. Yes, I missed mentioning in my post that members please disregard the obvious issues like focus/collimation etc (In my defence I have been imaging since 2006  ).
The exposure is a stack of 10 x 30s exposures. I will do further tests tonight on 300s x 10 Master darks and check.

 

 

 

[900SL]

8 minutes ago, astrosathya said:

 

Be aware that you will still have hot pixels in your 300s darks. These however calibrate out. Also, do investigate dithering when doing astrophotography

[astrosathya]

44 minutes ago, 900SL said:

Be aware that you will still have hot pixels in your 300s darks. These however calibrate out. Also, do investigate dithering when doing astrophotography

100%. I used to dither my DSLR and will also do it for this camera. I will test on Master darks tonight as the sky is likely to remain cloudy.

[astrosathya]

Guys, I just realised that 2 years of absolute hibernation does a lot of damage to an Astrophotographer. I am currently doing 180s test darks @ -10deg. 66% cooler. Each frame is virtually noise free. Nothing like yesterday. I will post a screenshot later when the test completes (phone is running ASIAIR App) but had to share this update.

In the screenshot, please ignore the massive light leaks I am facing. These are only tests I am doing. I will plug the leak which is from the mirror cell with appropriate media. 

Edited August 22 by astrosathya

[astrosathya]

Alright everyone, so here are the results. Please Ignore the Massive Light Leaks. I am going to fix it. But meanwhile please enjoy the gorgeous Master Dark File. 20 x 180s exposures. Smooth as butter, just the way I like it. 

One is obviously the full image screenshot and the second image is a screenshot of a zoomed in area of the Master Dark.