Resolution of astro CCD cameras

[petercoxphoto]

Hi folks -

I'm a professional landscape photographer and have recently started with astrophotography. I'm very much enjoying it, and there are definitely many important differences between the two types of photography - and I have an awful lot to learn.

For my landscape photography, I mainly use a medium format digital system - a Phase One IQ180, which offers 80Mp resolution. For when I need to work more quickly than this allows, I use a Canon 5D Mk II. This is the camera that I've been using so far in my astrophotography, paired with a 70-200mm f/2.8 lens. So for me, 21mp is my low resolution option.

I've had some success, and I do plan to buy a good astrograph in the medium term with the intent of doing mainly deep sky imaging. My problem is I'm not sure what to do about the camera.

It's down to the choice between DSLR and CCD. I already have the 5D, but it's not Baader modified, nor will it be as I need to use it in my daytime work also. So either way, for serious astro work I'll need a new camera.

I like the idea of the mono CCD cameras, and of course the reduced noise of a cooled CCD is an obvious advantage. With a mono CCD I should theoretically get about twice the effective resolution as it's not limited by the Bayer matrix. At least, in terrestrial imaging this is what I would expect - I'm assuming it is for astro work as well.

My problem is that it seems like 16mp is about the limit of astro CCD cameras, and for that you have to shell out a significant amount of money. Given that I make and sell large fine art prints of my images, and hope to do the same with my astro work, resolution is important to me. If a 16mp mono CCD can equate to a ~28mp one shot camera, that's not too bad.

I suppose my question is really - why are astro CCD cameras so low in resolution when compared to today's DSLR offerings? Is there a fundamental disconnect that I'm not seeing? Is resolution not really that important to astro imaging?

Any help would be welcome to ease my confusion. Particularly if any of you are making and selling large prints of your images, I'd be interested in your thoughts on resolution.

Thanks for your guidance!

Cheers,

Peter

[Merlin66]

"Resolution" as applied to Astronomy (and spectroscopy) is directly related to the plate scale of the telescope/ lens/ optics and the pixel size of the CCD detector.

The "best" resolution at the image plane will be close to twice the size of the pixel (in micron) - Shannon/ Nyquist sampling theory.

So for, say a camera with 5 micron pixels this would be 10micron.

OK, what does that mean re the astrophotographic image (or spectral image)?

The plate scale (arc sec/ pixel) will depend on the focal length of the telescope/ lens. If we have a telescope with a 1000mm fl the plate scale will be: 1 sec arc = 4.8micron. The resolution would then be around 9-10micron, and 2 arc sec.

That's the theory.

In real life the seeing conditions and atmospherics limit the size of the stars (and the detail see on the Sun/moon/planets etc).

Seeing of 2-4 arc sec would be typical for the UK.

The above example shows that the combination of the telescope/ camera is capable of recording twice as much detail as your average seeing conditions will allow. An even larger pixel camera may be a better solution!

Unless you intend imaging across the whole FOV ( and use flattners/ correctors etc etc etc) you don't always need the 80 Megapixel camera....the size of the planets will only cover a few pixels, same for galaxies and many star clusters...wide field nebula would be a full frame challenge...but just check out the issues/ problems/ concerns getting enough photons to smooth the noise (hours of exposure and more hours of processing!)

[IanL]

Seconded! Issues to consider:

- Getting a fully illuminated field is a significant issue for larger sensors. For an APS-C sized sensor (typical low-end DSLR) it is just about do-able with typical low-to-mid range amateur imaging scopes, but even then there is some vignetting at the edges of the frame. For a full-frame sensor (pro-DSLR) it is even more of an issue. You can correct vignetting by use of flat frames, but even so you want to start out with fairly good illumination across the whole sensor.

- Getting a corrected flat field is also an issue for larger sensors; egg-shaped stars around the edge of the frame and/or uneven focus from the centre to the outside are issues, and chromatic aberration (colour fringing) also comes in to play for refractors. The usual rule is that more money you spend on your scope/correctors the better the manufacturer will do in balancing all three, but the curve from affordable -> expensive -> "HOW MUCH?!!" is pretty steep.

- Signal to Noise Ratio is the key consideration when producing good deep-sky astro images (as you will discover). A general rule of thumb is that large pixels will give you a much higher SNR for a given amount of exposure time (or you can achieve the same by 'binning' smaller pixels, i.e. down-sampling and sacrificing resolution). Matching pixel-size to the imaging equipment and target are the name of the game. The ideal pixel size for a long focal length, slow (high f-number) scope suitable for imaging small galaxies is not the same as for a short focal length, fast scope used for imaging extended nebulae.

All in all, resolution (of the camera) is generally a secondary consideration when producing good astro-images compared to the problems above. If your aim is to produce big images to create large prints for sale, of course you need sufficient resolution to enable them to be enlarged and still look goof. There is of course a simple (if time consuming answer), which is to mosaic multiple frames together to cover more of the sky. Unlike daytime photography, the subjects of astro-imaging are (almost invariably) unchanging so it is entirely feasible to spend several months (or even several years) capturing enough data across many mosaic panels to create the large/spectacular images we all love so much.

Doubtless some of our obsessive brethren will emerge from the woodwork in due course to confirm that latter point

[GlassWalker]

Personally I've always wondered why they don't take a DSLR sensor, remove the colour filter, stick cooling on it and sell that for astro imaging? But until someone does, we have to make do with what's available.

Regarding the impact of bayer pattern sensors and alternately mono imaging, I'd argue for deep sky astrophotography, the effective resolution of the mono sensor approach is much greater than it may first appear. H-alpha emissions are common in many nebula, and sit in the deep red region. Red detectors make up only 1/4 of a bayer pattern sensor, so I would argue in this application, it would be valid to say the mono sensor is equivalent to a colour sensor of 4x the MP count.

Take a Kodak KAF 8300 based CCD for example as a high-ish MP yet affordable astro imager. It is 8.3MP, but for subjects featuring lots of red, it would be equivalent to a bayer pattern sensor of around 33MP. Ok, not all deep sky objects are H-alpha dominant, for example, there are numerous galaxies which are more broad spectrum. Arguably the mono sensor benefit is reduced here, but in my opinion we're still looking at a factor of 2x.

[dph1nm]

Personally I've always wondered why they don't take a DSLR sensor, remove the colour filter, stick cooling on it and sell that for astro imaging

Err - isn't that basically what most commercial amateur astro CCD cameras are?

NigelM

[Tom OD]

Great explainations here thanks guys.

Tom.

[r3i]

Personally I've always wondered why they don't take a DSLR sensor, remove the colour filter, stick cooling on it and sell that for astro imaging?

They have done with the Sony ICX 453AQ SuperHAD CCD found in the Starlight Xpress SXVF-M25C and the QHY8/QHY8 Pro - this chip was in some Nikon DSLRs like the D50 & D70.

[GlassWalker]

Err - isn't that basically what most commercial amateur astro CCD cameras are?

DLSR sensors are pretty much all CMOS now. They moved away from CCD years ago which helps a LOT with noise for "normal" photography in lower light conditions. I think in theory, CCDs have the overall performance edge in astro use, but CMOS might not be that far off. Potentially it could be a lot cheaper if leveraged off mass market DSLR sensor developments.

[GlassWalker]

They have done with the Sony ICX 453AQ SuperHAD CCD found in the Starlight Xpress SXVF-M25C and the QHY8/QHY8 Pro - this chip was in some Nikon DSLRs like the D50 & D70.

Ok, I wasn't aware of that, but I was thinking of something more recent! If you look at the output from cameras of that time, and compare it with what you can get now, things have moved on a lot.

[ollypenrice]

Some good answers up there ^^ .

I always groan when the word 'resolution' comes up because it has far too many meanings in astro imaging. They key things for me are these;

1) Because of the seeing limit and the low signal to noise ratio smaller pixels do not necessarily increase visible detail in astro images. As stated above they may diminish it. And smaller pixels may well give larger stars.

2) Before you can get anywhere near to realising the pixel scale/optical resolution of your scope and camera you need to get enough signal. Working at F3.9 I think I can get signal acceptable for the brighter objects in 10 hours or so. Working at F6.8 and F7 I would put that at more like 20 hours. It would be longer, but the long FL targets in our F7 scopes tend not to need so much narrowband input.

Mosaics are the way forward but they are long, long projects. I spent 60 hours on the Orion constellation. (6x10 hours.) Above all a fast F ratio is the thing with a tandem scope rig being another option.

And in my view CCD cameras are way ahead of DSLRs whatever is done to them. They are easier to use, too.

Olly

[IanL]

Canon develop their own CMOS sensors and don't resell them to third parties. Their only forays into using these sensors for astro-imaging is to put in a different IR blocking filter to increase red/Ha sensitivity. I suspect that even something as minor substituting a different filter for small production runs is pretty disruptive/expensive for them to do (hence the high price of the 60Da and the 20Da before it).

To make these sensors properly viable for astro cameras would be a major undertaking (re-writing all the firmware to eliminate the daylight processing/optimisation and replace it with something closer to the direct from chip output you get from a CCD for one thing). Even if Canon were willing to supply their CMOS sensors (and proprietary secrets needed to build a third party camera using them), I very much doubt the final price would be any different than that of any other large chip astronomical CCD camera.

The reason you get huge sensors in DSLRs so much cheaper than big sensors in Astro CCDs is down to economies of scale, i.e. R&D, manufacturing, markerting/sales and support costs spread over the huge DSLR market. Making a different line of astro cameras would involve a lot of the same costs spread over a smaller market.

[petercoxphoto]

Hi folks -

Thanks for the very helpful responses. Plenty of food for thought here - I'm going to have to and digest it now. Clearly there are significantly different requirements here. My interest would be for both larger nebulae and galaxy imaging, so it sounds almost as though to do it properly I'd need two scopes, and possibly two cameras? (On top of guidescopes and cameras for those purposes...).

Is that a correct statement?

Cheers,

Peter

[dph1nm]

DLSR sensors are pretty much all CMOS now.

The Kodak 8300, much used in astro cameras, was original used in Olympus DSLR cameras. I had the impression (and I might be wrong) that most astro cameras were (are?) based on old CCDs originally developed for consumer cameras.

NigelM

[opticalpath]

I think another important difference is in the camera electronics. Daytime photography: high signal, short exposure, noise not too significant. Deep sky imaging: very low signal, long exposures, noise level critical. The cooling and signal processing electronics in astro cameras is designed to keep noise to a minimum during long exposures. (Notwithstanding very good performance of modern DSLRs in that respect.)

Adrian

[Tom OD]

Peter,

I think you can see now why I was suggesting that you start with the cameras you have, to get a feel for the photography. I image at 530mm,so for me the only galaxies I can really have a go at, are Andromeda and M33. Well I guess a widefield of Markarians chain. I would like to go over 1.5m for galaxies but as you ,thats another scope, not necessarily another camera. A long f/l scope with a shorter one piggy backed is a good set up. You can switch the imaging camera between the two to guide. You need a good big mount then though. I would stay at the lower f/l lenght for the moment to get the hang of the processing and euipment before considering the longer f/l's.

T.

[Tim]

Hi Peter,

Great post

Don't forget with Astro imaging you can up sample your images very effectively using 2 x or 3 x drizzle techniques.

What sort of photo's are you hoping to take? Widefield shots or close up and detailed? As explained above, you can easily have too much camera. The focal length you decide on will help you best determine the camera that is right for you.

How are your skies? In dark sky areas OSC cameras are a great option, and can save a lot of mucking about, but for more sensitivity and for narrowband filter work, then a mono camera is the choice. Personally I have and use both to suit the skies and the target.

Another way of increasing 'resolution' is to use several images in a mosaic.

As you say, astrophotography makes you look at things in a whole new light.

This website and program may help you a bit with a feel for the differences of focal lengths and actual imaging resolution in arcsec/pixel - http://www.newastro.com - Look for the CCD calculator and download the extra images.

You can even put your own camera details in with a combination of telescopes and lenses to see the difference in a graphical presentation.

Hope that helps a bit.

Tim

[swag72]

My interest would be for both larger nebulae and galaxy imaging, so it sounds almost as though to do it properly I'd need two scopes, and possibly two cameras? (On top of guidescopes and cameras for those purposes...).

I kind of agree with this statement - Galxies (apart from the normal suspects,ie M31) benefit from a focal length of over 1m. Larger nebula work well on less than 500mm - I think that unless you want multiple scopes (I decided that I didn't) then you need to get to a stage where your targets for imaging are compatible with your scope. Many people use an ED80 refractor and a small chip CCD, such as the Atik 314L+, to great effect. It is hard to work out where your main interest lies - Mine worked out as being an expensive learning curve, but I have happily got there in the end.

[Gina]

ATM I have 2 sizes of camera with ED80 scope and several lenses (kit in signature). That covers wide field and medium DSOs - should keep me occupied for some time But later I expect I shall want to image the smaller DSO such as galaxies, then the 600mm ED80 and the Atik 314L+ won't cover a narrow enough angle and I guess I shall be looking at a bigger scope such as the MN190.

[Uranium235]

Ive been experimenting with both systems running in tandem, a CCD for luminance data, and a DSLR for colour data on identical telescopes (ED80). Still early days, but im starting to get the hang of it - and for bright targets and galaxies it works quite well. However for narrowband work, 2 CCDs would be better.

Theres no need for a massive sensor if youre looking for galaxies since all bar one (M31) will easily fit on to a chip the size of a Sony 285 (even at 1000mm FL). To get some of the larger nebulae, you can either reduce your FL (the cheap option), do mosaics (the time-consuming option), or get a bigger chip - the Atik 460EX would fit the bill nicely but that will set you back over 2k.

Here is an example of both CCD and DSLR in harmony, taken 4 nights ago in one session - if i'd used individual filters, it wouldnt even be half finished:

[Gina]

That's a superb image I think I shall have to try that. I have adapters coming to let me connect my 314L+ and EFW2 to Pentax/Praktica mount lenses - of which I have a good selection of focal lengths. 35, 55, 135, 200, 300mm. Some are genuine Asahi SMC Takumar and excellent lenses. If I use 314L+ with lens and 1100D (modded and cooled) on ED80, I think I might have a useful setup. I could arrange that both systems have similar coverage.