First astro camera - 269 or 294 sensor - sampling confusion

[Milesy303]

Hi all.

I’m currently looking at both sensors to use with my GT81 scope for a few months but want to go to a longer focal length for after the summer. I have been looking at the same targets at the same resolution for quite a few years with my DSLR and my primary objective is getting into smaller targets and more detail.

With the 269 I’ll be over sampled when I go to the longer focal length but there is no amp glow to deal with.

The 294 seems an older technology but maybe bet better suited to upgrading after the summer ?

To be used under "uk seeing" which I think might be "ok"

My head hurts with all the sampling and seeing.

 

Chris

[smudgeball]

Hi, I've used my 269 with a Megrez 72 (current) and GT102 (currently stored)

I agree that all the online calcs and "seeing" etc can be too much.

I also think the numbers you get are based on ideal conditions (what are they ?)

Certainly the 269 gives very clean subs, I do calibrate but mostly for the dusty obsy I have 🙂

Neil

[Milesy303]

29 minutes ago, smudgeball said:

Hi, I've used my 269 with a Megrez 72 (current) and GT102 (currently stored)

I agree that all the online calcs and "seeing" etc can be too much.

I also think the numbers you get are based on ideal conditions (what are they ?)

Certainly the 269 gives very clean subs, I do calibrate but mostly for the dusty obsy I have 🙂

Neil

Thanks Neil, how did it serve up with the GT102? Do you take darks?

[Adam J]

52 minutes ago, Milesy303 said:

Hi all.

I’m currently looking at both sensors to use with my GT81 scope for a few months but want to go to a longer focal length for after the summer. I have been looking at the same targets at the same resolution for quite a few years with my DSLR and my primary objective is getting into smaller targets and more detail.

With the 269 I’ll be over sampled when I go to the longer focal length but there is no amp glow to deal with.

The 294 seems an older technology but maybe bet better suited to upgrading after the summer ?

To be used under "uk seeing" which I think might be "ok"

My head hurts with all the sampling and seeing.

 

Chris

ASI533mc pro is the answer in my opinion. After that the ASI294mc Pro but its a little trickier to calibrate, there are a few long threads on SGL covering this topic. 

I don't like what i see from the 269 as there seems to be allot of walking noise in most of the images i see. Guiding and dithering would be essential in my opinion. 

Adam 

Edited February 7, 2022 by Adam J

[Milesy303]

25 minutes ago, Adam J said:

ASI533mc pro is the answer in my opinion. After that the ASI294mc Pro but its a little trickier to calibrate, there are a few long threads on SGL covering this topic. 

I don't like what i see from the 269 as there seems to be allot of walking noise in most of the images i see. Guiding and dithering would be essential in my opinion. 

Adam 

I am looking at Altair, the 533 seems quite a bit cheaper than the 269 and the 294. I assumed it was less of a camera.

[smudgeball]

34 minutes ago, Milesy303 said:

Thanks Neil, how did it serve up with the GT102? Do you take darks?

I take darks but there is zero glow from my example.

I find noise well controlled on it.

I did consider the 533 but wasn't sure about a "square" sensor.

I'm not sure if price point necessarily equates to quality in that regard?

From my 102

[drjolo]

2 hours ago, Milesy303 said:

With the 269 I’ll be over sampled when I go to the longer focal length but there is no amp glow to deal with.

I would not worry with modern CMOSes oversampling at all. 294 is larger than 269 and also larger than 533 - and that is the quality that is often ignored, but for astrophotography that is quite fundamental. Together with sensitivity (quantum efficiency). There were quite many nights that I missed larger sensor. Harder calibration or dithering requirement was never a pain for me    

Edited February 7, 2022 by drjolo

[vlaiv]

9 minutes ago, drjolo said:

I would not worry with modern CMOSes oversampling at all

I worry about that all the time

1 hour ago, Milesy303 said:

I’m currently looking at both sensors to use with my GT81 scope for a few months but want to go to a longer focal length for after the summer. I have been looking at the same targets at the same resolution for quite a few years with my DSLR and my primary objective is getting into smaller targets and more detail.

Do you have any idea of what sort of working resolution you want to achieve?

[Adam J]

56 minutes ago, Milesy303 said:

I am looking at Altair, the 533 seems quite a bit cheaper than the 269 and the 294. I assumed it was less of a camera.

Its a smaller sensor so that is why its cheaper, but probably a higher quality sensor. So depends on what you want to take a picture of, if your targets fit on the 533 then its likely the best bet. If you want a wider view then I would say the 294mc pro as its got a higher QE, but calibration while possible is not so reliable. 

I don't think that under-sampling is going to be an issue for you. Most of the time you will not resolve detail to the pixel level in any case, especially if you dont have good SNR and you have to apply noise reduction. Better to start out with higher SNR from a larger pixel higher QE sensor. 

Adam

Edited February 7, 2022 by Adam J

[drjolo]

18 minutes ago, vlaiv said:

I worry about that all the time

Why do you? The only problem with oversampling is adding the read noise to the signal. And for CMOS cameras that is not a big problem, unless you want to do narrow bands with OSC camera. 

[Adam J]

34 minutes ago, drjolo said:

I would not worry with modern CMOSes oversampling at all. 269 is larger than 294 and much larger than 533 - and that is the quality that is often ignored, but for astrophotography that is quite fundamental. Together with sensitivity (quantum efficiency). There were quite many nights that I missed larger sensor. Harder calibration or dithering requirement was never a pain for me    

The 269 is not larger then the 294.

[drjolo]

2 minutes ago, Adam J said:

The 269 is not larger then the 294.

Yes, you are right, my mistake. The information at the page https://www.altairastro.com/altair-hypercam-269c-colour-camera---tec-cooled-1097-p.asp is misleading

"5280x3956 active 3.3 micron pixels in a 20.49 mm × 17.83 mm array. Physical sensor size, 4/3rds" 21.77mm diagonal."

[vlaiv]

3 minutes ago, drjolo said:

Why do you? The only problem with oversampling is adding the read noise to the signal. And for CMOS cameras that is not a big problem, unless you want to do narrow bands with OSC camera. 

It considerably lowers SNR without any real benefit.

Since aperture does not change - setup gathers same amount of photons per unit time - but spreads those photons over more pixels. Each pixel then receives less photons in unit time - so system is slower and achieves lower SNR in set imaging time - again without any benefit in terms of detail since we are over sampling.

With small pixels of modern cmos cameras it is very easy to over sample.

[Adam J]

Just now, vlaiv said:

It considerably lowers SNR without any real benefit.

Since aperture does not change - setup gathers same amount of photons per unit time - but spreads those photons over more pixels. Each pixel then receives less photons in unit time - so system is slower and achieves lower SNR in set imaging time - again without any benefit in terms of detail since we are over sampling.

With small pixels of modern cmos cameras it is very easy to over sample.

One thing to think about is if the lower QE 269 may be 2x2 binned and get better SNR than a unbinned 294 with higher QE / lower read noise. Suspect it would be close but am at work so dont have time to run any numbers. 

Adam 

[vlaiv]

5 minutes ago, Adam J said:

One thing to think about is if the lower QE 269 may be 2x2 binned and get better SNR than a unbinned 294 with higher QE / lower read noise. Suspect it would be close but am at work so dont have time to run any numbers. 

Adam 

I think camera choice needs to be paired with telescope choice and of course - depending on mount.

First step should be determining desired working resolution. Then we can discuss if that is realistic and under what circumstances.

In the end - we can see what camera and binning combination will get us closest to that.

[drjolo]

15 minutes ago, vlaiv said:

It considerably lowers SNR without any real benefit.

I may not agree, unless you define "considerably".  I just checked a 180s luminance frame captured with f/10 telescope and QHY163M camera. Background noise is 167ADU, while bias noise at the same gain is 20ADU. Background noise with 25ADU read noise would be about 167.7ADU, and with 15ADU read noise would be around 166.5ADU. I would not call it a significant difference. 

15 minutes ago, vlaiv said:

Since aperture does not change - setup gathers same amount of photons per unit time - but spreads those photons over more pixels. Each pixel then receives less photons in unit time - so system is slower and achieves lower SNR in set imaging time - again without any benefit in terms of detail since we are over sampling.

It is based on assumption, that smaller pixel has the same readout noise as larger (and other noise sources are insignificant). But that is rarely true. If the read noise of the smaller pixel is smaller proportionally to the pixel area difference, then there is no loss in SNR. 

Edited February 7, 2022 by drjolo

[Adam J]

14 minutes ago, vlaiv said:

I think camera choice needs to be paired with telescope choice and of course - depending on mount.

First step should be determining desired working resolution. Then we can discuss if that is realistic and under what circumstances.

In the end - we can see what camera and binning combination will get us closest to that.

We know its a GT81 from OP, going to assume he can get an RMS of around 1 arcsecond unless he is under-mounted. 

As above and more so with OSC than Mono i think that even if you can get better theoretical "working resolution"  by choosing a smaller pixel camera then you still need sufficient SNR to take advantage of that available resolution. In the majority of cases it has been my experience that resolved detail is limited by SNR and not pixel scale. 

I have seen renditions of the vail SNR taken with large aperture long focal length scopes at 1.5arcseconds per pixel that have less resolved detail than my ASKAR FMA180 at 4.5arcseconds per pixel,  just because my image has better SNR and I hence have applied greatly less or no noise reduction. Its an extreme but true example. 

Adam 

[vlaiv]

2 minutes ago, Adam J said:

We know its a GT81 from OP, going to assume he can get an RMS of around 1 arcsecond unless he is under-mounted. 

As far as I understood OP is looking for longer FL scope to pair with new camera?

[Adam J]

2 hours ago, vlaiv said:

As far as I understood OP is looking for longer FL scope to pair with new camera?

Yes so I see, well for what its worth having reviewed many images of NGC4565, I have seen very few (taken from UK close to sea level) that have much more resolved detail than what I managed at 1.4acrseconds per pixel at 550mm focal length. 

With the GT81 and a 533mc it comes out at around 1.65 the 269 would be close to the Daws limit of ~1.4, you might be able to go lower than 1.4 but to be honest I think you would have to start being very selective about what subs you accept into the stack or be very lucky with seeing. So for a larger scope to image smaller targets I would, go no further than say a 533 and a scope with around 700mm focal length but for me in the UK that is unlikely to get you much further than I did below so i see no real harm in going to the larger pixels of the 294. All in all though i think its best to go mono for high resolution imaging as you get more detail at a given focal length and pixel size without the colour matrix.

Adam

Edited February 7, 2022 by Adam J

[vlaiv]

2 hours ago, Adam J said:

I have seen very few (taken from UK close to sea level) that have much more resolved detail than what I managed at 1.4acrseconds per pixel at 550mm focal length. 

Out of interest, what is average FWHM of stars in the image in linear data?

[Adam J]

47 minutes ago, vlaiv said:

Out of interest, what is average FWHM of stars in the image in linear data?

Not sure about FWHM but my Esprit 100 generally produces a HFR of 1.7-1.8 in good conditions according to NINA with a ASI1600mm pro and at 1.4 arc-seconds per pixel. Dont have access to the data as that is archived on a removable HDD. Might have time to dig it out if you really want me to. Best ever HFR = 1.5-1.6

If i am converting correctly i could still be over sampled by as much as 1/3rd.

Adam 

Edited February 7, 2022 by Adam J

[vlaiv]

2 minutes ago, Adam J said:

Not sure about FWHM but my Esprit 100 generally produces a HFR of 1.7-1.8 in good conditions according to NINA with a ASI1600mm pro and at 1.4 arc-seconds per pixel. Dont have access to the data as that is archived on a removable HDD. Might have time to dig it out if you really want me to. 

Adam 

I think it would be interesting to see and compare with FWHM approximation.

If it's too much hassle - then don't bother, but if not, I would be interested in checking it out.

We can go by HFR for the time being. If it is 1.7-1.8 - that would put HFD at 3.4-3.6" (HFR is half flux radius while HFD is half flux diameter). For perfect Gaussian HFD is the same as FWHM, so it is safe to say that your images have 3.4-3.6" FWHM star profiles.

That is what I would expect from 100mm of aperture under "regular conditions" (seeing 2" and guiding of about 1" RMS - actual calculation gives ~3.3" FWHM).

By the way - that is 2"/px sampling rate rather than 1.4"/px

Btw, one of these two has been reduced to 66.7% in size and then resized up to original size (150%).

 

[Adam J]

4 minutes ago, vlaiv said:

I think it would be interesting to see and compare with FWHM approximation.

If it's too much hassle - then don't bother, but if not, I would be interested in checking it out.

We can go by HFR for the time being. If it is 1.7-1.8 - that would put HFD at 3.4-3.6" (HFR is half flux radius while HFD is half flux diameter). For perfect Gaussian HFD is the same as FWHM, so it is safe to say that your images have 3.4-3.6" FWHM star profiles.

That is what I would expect from 100mm of aperture under "regular conditions" (seeing 2" and guiding of about 1" RMS - actual calculation gives ~3.3" FWHM).

By the way - that is 2"/px sampling rate rather than 1.4"/px

Btw, one of these two has been reduced to 66.7% in size and then resized up to original size (150%).

 

The resized one is on the left, its subtle but I can see a loss in detail / slight increase in softness. I probably resized it to 2 arcseconds myself, it was a while back. 

Edited February 7, 2022 by Adam J

[vlaiv]

1 minute ago, Adam J said:

The resized one is on the left, its subtle but I can see a loss in detail / slight increase in softness. 

Check that left against original.

(hint - I think I left sharpen when resampling turned on )

[Adam J]

2 minutes ago, vlaiv said:

Check that left against original.

(hint - I think I left sharpen when resampling turned on )

Probably need to sharpen the original to compare then.