Jump to content

NLC-Banner.thumb.jpg.acb5ba835b9e8bf0718b90539633017d.jpg

To CMOS or not???


petevasey
 Share

Recommended Posts

On 01/11/2021 at 06:54, newbie alert said:

So carry on wondering around your park at night, thinking about selling your mount for a low cost version and panicking about Mr musk and friends...

unfortunately we have an infestation of men with guns looking to bag about 80 bodies... so the night park wanderings are off for a while.

 

On 01/11/2021 at 12:37, DaveS said:

The QHY 410C looks to be a good match for my Moravian G3 16200. Use the G3 for Lum and NB, and the QHY for RGB. Haha, upper case 3 is £, so nearly typed G£ 16200 😆. Appropriate really.

I think this is the answer to anyone with ccd thinking of jumping into cmos, keep the ccd for mono to grab the detail of the image and  get a cmos osc to quickly grab RGB to colour that detail.

Edited by dmki
Link to comment
Share on other sites

5 hours ago, dmki said:

unfortunately we have an infestation of men with guns looking to bag about 80 bodies... so the night park wanderings are off for a while.

 

I think this is the answer to anyone with ccd thinking of jumping into cmos, keep the ccd for mono to grab the detail of the image and  get a cmos osc to quickly grab RGB to colour that detail.

Unclear to me why you think a CCD would be better to grab the details, especially compared to a modern mono CMOS, considering the much higher read noise, usually lower QE, dead pixels and columns that affects the CCD.

Link to comment
Share on other sites

8 minutes ago, DaveS said:

I just wish someone would make a BSI CMOS sensor equivalent to the 16200. APS "H" format and 6 micron pixels. Could easily have a FWC greater than 100,000 like the Sony IMX 410.

But more megapixels = better camera, right? /s

Megapixels of a camera is like a department store telescope advertising "100x magnification!" = marketing fluff, where a novice sees a big number and likes it better than the small number.

Link to comment
Share on other sites

2 minutes ago, DaveS said:

16 megapixels. On my ODK that gives a still ridiculous 0.61"/px

These crazy high megapixel sensors like the IMX 455 only work if you bin the pixels, which can only be done in software.

Or short focal length scopes and camera lenses but even then the pixel sizes are still not small enough.

Alan

Link to comment
Share on other sites

12 minutes ago, DaveS said:

16 megapixels. On my ODK that gives a still ridiculous 0.61"/px

These crazy high megapixel sensors like the IMX 455 only work if you bin the pixels, which can only be done in software.

That pixel size is perfect for something like a RASA or Epsilon, and even 1 m FL refractors, and what is the problem with binning if you have a longer FL? Nothing wrong with software binning as far as I know, and the great thing is that you can decide after collecting the data if you want to do it or not. That is not an option with CCD. A camera with small pixels that can be binned is much more versatile than one where you are stuck with big pixels, especially with the new CMOS that have quite a lot of well depth.

  • Like 2
Link to comment
Share on other sites

9 hours ago, dmki said:

 

I think this is the answer to anyone with ccd thinking of jumping into cmos, keep the ccd for mono to grab the detail of the image and  get a cmos osc to quickly grab RGB to colour that detail.

I have been doing exactly that on my dual refractor rig, capturing L with a KAF8300 CCD and a 0.77 focal reducer, and RGB with a QHY268 OSC, this arrangement gives almost the same FOV. 

It works, but I’m inclined to agree with Goran, I think the Lum data would be better with a mono version of the CMOS camera. 
 

Based on a study of the Astro classifieds, the window has closed now I think, on selling on the CCD at a reasonable price to help fund another CMOS camera, so I’ll hang on to it for now. It will probably go in my Museum of Imaging Artefacts, alongside my SBIG ST4 guide camera, which incidentally, originally cost as much as a decent present day CMOS camera.☺️

  • Like 1
Link to comment
Share on other sites

5 hours ago, gorann said:

Unclear to me why you think a CCD would be better to grab the details, especially compared to a modern mono CMOS, considering the much higher read noise, usually lower QE, dead pixels and columns that affects the CCD.

I am not.... i am suggesting that luminescence is where we perceive detail, so keep your existing mono ccd to capture luminescence and get a cmos osc to capture rgb... reducing your image capture times.

Link to comment
Share on other sites

6 hours ago, dmki said:

I am not.... i am suggesting that luminescence is where we perceive detail, so keep your existing mono ccd to capture luminescence and get a cmos osc to capture rgb... reducing your image capture times.

Are you saying swap your cameras between sessions?  I can’t think of anything more likely to create headaches to be honest! 

Link to comment
Share on other sites

1 hour ago, Ags said:

Well, you could use both cameras in a dual scope setup.

51 minutes ago, tomato said:

Yes, see my post higher up, but a dual CMOS set up is the way to though, if you have the budget.

If you are using a fast scope, and have the tilt/spacing sorted, swapping cameras would be a nightmare IMHO!  Two scopes is definitely the way to go regardless of CCD or CMOS

Link to comment
Share on other sites

Totally agree, swapping filters in a session has an engineered solution (wheel or drawer), but cameras is a whole different ball game. 
 

Rotating turret anyone? Imagine the size of the focuser to carry that!

  • Haha 1
Link to comment
Share on other sites

The QHY 410 makes sense *to me* financially, as the QHY 600 is £1000 more without adding filters or wheel. And I'm not sure about binning colour cameras, I have an idea that it has to be done 3x3?

  • Like 1
Link to comment
Share on other sites

20 minutes ago, DaveS said:

The thought of *two* ODK 12s on the DDM 85 *is* the stuff of nightmares, one is quite enough (Or more than enough) though I think the DDM 85 would *just about* carry them.

Sure you could just buy a second DDM 85 😂

Link to comment
Share on other sites

46 minutes ago, DaveS said:

The QHY 410 makes sense *to me* financially, as the QHY 600 is £1000 more without adding filters or wheel. And I'm not sure about binning colour cameras, I have an idea that it has to be done 3x3?

I have been told that using superpixel when debayering in PI is effectively a 2x2 binning and I have tried it with ASI2600MC and it works. So binning is possible with colour CMOS. I am sure @vlaiv may know even better ways of doing it.

PS. Why not a QHY600/ASI6200 colour camera? That is 16 bit - I thought QHY410 was 14 bit.

Edited by gorann
Link to comment
Share on other sites

20 minutes ago, gorann said:

I have been told that using superpixel when debayering in PI is effectively a 2x2 binning and I have tried it with ASI2600MC and it works. So binning is possible with colour CMOS. I am sure @vlaiv may know even better ways of doing it.

PS. Why not a QHY600/ASI6200 colour camera? That is 16 bit - I thought QHY410 was 14 bit.

Yes, super-pixel mode rings a bell, though I don't have PI. As for the cameras, yes the QHY 410* is 14 bit which may be a problem, but so is the £1000 extra just for a bare QHY 600C.

But all this is somewhat academic since it will be well into next year before I can even think of a new camera.

 

*Not to be confused with the QHY 411 which is a whole 'nother ball game and into second mortgage territory.

Edited by DaveS
Link to comment
Share on other sites

11 minutes ago, gorann said:

I have been told that using superpixel when debayering in PI is effectively a 2x2 binning and I have tried it with ASI2600MC and it works.

Here we have to be careful what we mean when we say OSC binning. OSC binning is possible in every sense of the word binning - but it is a bit different to mono binning.

We expect two major things to happen when we bin:

1. we expect resolution / sampling rate to go down

2. we expect SNR to go up

Both of above happen for mono data predictably - if we bin with certain factor - say N (x2, x3, x4 ...), then - resolution goes down by factor N and SNR improves by factor N for mono data.

For OSC data - it is not quite like that, because OSC data is "sparse" data to begin with. We think that we sample at certain resolution / sampling rate with it, and that our sensor has "that much pixels" - but with OSC cameras that is not quite true.

We only have 1/4 of total pixels being red, and 1/4 of total pixels being blue and 1/2 of total pixels being green (for RGGB type Bayer matrix - there are other types that don't use RGB or use it differently - but all astro OSC cameras use some sort of RGB with two green pixels).

So for OSC things are as follows:

1. If we use interpolation debayering - one that gives us "regular" resolution / pixel count for sensor - and we bin - following will happen:

- resolution / sampling rate will go down as N

- SNR won't improve as N. This is because SNR improvement expect "pure pixels" (non correlated), but when we interpolate - we introduce pixel correlation - 3/4 of red pixels are in fact just derived from that 1/4, so they are not valid "measurements". This is a bit like - copying some of subs and then stacking that and expecting SNR to improve - that won't happen as we don't have new data - just several copies of old/existing data

2. If we use split / super pixel debayering - one that gives us only half of resolution of the sensor (or 1/4 of pixel count)

- resolution / sampling rate will go down as 2*N - this is because we "acknowledge" that our sensor is in fact "sparse" and that it is already sampling at half the resolution for each color (here we treat green as two "colors" - G1 and G2 - but we will stack them on the same stack so green will be slightly better than other two colors in terms of SNR - which is good as it carries the most luminance information).

- SNR improvement will be as N

------------------------

With above - we can see that super pixel mode debayering is not really 100% like binning. It does reduce sampling rate, but it does not improve SNR over interpolation debayering as much as binning does. In fact - it only improves green channel as it "stacks" two green components, but red and blue remain as they are.

28 minutes ago, gorann said:

PS. Why not a QHY600/ASI6200 colour camera? That is 16 bit - I thought QHY410 was 14 bit.

You really should not get too hung up on bit count - it is really not important if you handle your data properly. Many people use ASI1600 and produce excellent results - and it is only 12bit camera.

Each time we stack - we increase bit depth of final stack, so even 12bit camera can easily produce 20+ bit precision image.

  • Like 3
  • Thanks 1
Link to comment
Share on other sites

20 minutes ago, vlaiv said:

You really should not get too hung up on bit count - it is really not important if you handle your data properly. Many people use ASI1600 and produce excellent results - and it is only 12bit camera.

Each time we stack - we increase bit depth of final stack, so even 12bit camera can easily produce 20+ bit precision image.

This is particularly applicable with many short sub-exposures compared to fewer longer ones. So it's another win for CMOS.

BTW the extra 8 bits of resolution would require a minimum of 256 subs - that's a lot of storage and CPU demand.

Link to comment
Share on other sites

On 30/10/2021 at 17:48, petevasey said:

Hi, all,

For almost 12 years now I've been imaging with my QSI683 wsg camera.  8-way filter wheel, built in off axis guider, good sized CCD chip.  And of course getting good results.  But friends of mine are using the ASI533 one-shot colour CMOS camera, and rave about it!  Very sensitive and with the appropriate multi band filter appears to take good narrow band type images.  So I've been browsing similar cameras, and believe the ASI 294MC Pro will nicely fit the bill - almost identical chip size so no problems with flat fields, very sensitive and not a huge pixel count.  I believe there may be some amp glow, but that's easily taken care of with darks.  I'll be using Windows XP, but tests with one of the 533 cameras were successful, so hopefully the 294 will also run ok.  I'd probably try and persuade a supplier to do me a sale or return so I could confirm that all is well.

BUT...  Although a one-shot colour is very tempting - it would seem that with the high sensitivity these cameras can considerably reduce total imaging time even though binning is not possible, nevertheless will a change really be worth it?  Any comments?

Cheers,

Peter

Hi Pete, I own an Atik383 same sensor as your QSI and ive owned a ZWO533MC for 11 months, with good cameras that compliment each other well.

Personally I would avoid the 294MC as the effects of starburst amp glow are a complete shock to the system and contrary  to popular belief just subtracting darks doesn't completely deal with the effects of amp glow, yes the the glow disappears but what remains in that area will be greater noise, Steve Chambers from Atik has a great simple explanation and demonstration in the video below using the Atik Horizon cmos.

Now for me I couldn't see any point in buying a camera that has such strong amp glow when the latest generation of ZERO amp glow sensors are available, which is why I went for the 533. If I had more money its larger brother the 2600. It Just seems crazy to go from a CCD sensor that is so clean and glow free to a sensor that has massive glow. Also the 294's can suffer with strange coloured blotches in the Flats and lights, especially when using dual narrowband filters. 

The 533 and 2600 are as close to CCD files as ive seen from a CMOS.  I can take 15 min subs with the 533 and still not see any Amp glow. ive used the hundreds of short subs method with it and also tried my standard CCD exposure times and cant see much difference in the final stack, so I normally go for longer subs to minimise storage and processing time ( as im the laziest imager ever ). in fact for me the short subs have a rarely talked about draw back of fixed pattern banding noise that isn't present in longer subs, I recommend not going shorter than 30-60 secs if you want to see clean subs. Nearly every discussion about CMOS concentrates on the low read noise allows shorter subs to give an advantage and while in theory this is true, ive found in real word use the lack of dark current and fixed pattern noise is far more important to me. The reason for this is that even comparably high read noise of the noisest CCD cameras will easily get swamped in even the slightest light pollution by shot noise. While fixed pattern noise on the other hand does not. So if I take hundreds of 10 sec shots that all have a banding in them, when I come to stack them the pattern will actually become more evident rather than decrease hence the absolute need for dark frames to compensate ( also shown in Steves video ). If I take subs of over 60 secs each they show no discernible banding at all, hence I never use darks with my cmos camera and have never seen any benefit from using them, as long as I use long enough subs. So I simply dither my long subs and get perfectly clean results.  In short read noise isn't important to me at all, even in narrowband there is enough light pollution getting through to swamp even the highest read noise.

Finally id like to Add that the claimed higher sensitivity of cmos cameras never really manifests itself to me in the real world except when doing high frame rate planetary imaging.

Prior to the 533osc I owned an ATik428osc and I have pretty much identical subs taken with both and for the same sub lengths there was very similar signal picked up with both on faint targets, I really cant say one is better than the other, actually the only reason I switched was that the 533 is a bigger chip for the same price. My supposedly ssllooww Kodak Mono sensor  is still much faster than either.

I will probably get loads of hate now, but I prefer to go on my own real world tests than popular opinion.

Id recommend the 533 or 2600 over the 294 any day.

Lee

 

  • Like 3
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.