How can I get Increased resolution for lunar imaging?

[andyrawlins]

Hi All, I'm looking for advice on getting more resolution when taking photos, primarily of the moon.

I'm using an 8" Newtonian Reflector, 1200mm focal length, with a full frame Canon 5D III and 2x teleconverter.  This gives me nice sharp images that almost fill the frame but, even with 22.3 megapixels, I get limited 'zoom in' on photos before pixilation sets in.   

I have also been using APT and EOS Camera Movie Record to record the live view at 5x.  This also gives good results but as far as I know this just optimises the resolution (and allows stacking).  It can't actually increase the native resolution of the system surely?

I have also tried with a 2x Barlow, with and without the teleconverter, but this requires the use of extension tubes which leaves a heavy camera hanging on a long stalk of tubes that is neither stable nor safe for my camera.  My SLR adaptor allows me to insert objectives but this gives poor results, from uneven lighting to images severely stretched from the centre.  An adjustable length adaptor might be more successful, I don't know.  

Another possibility would be a dedicated astrophotography camera such as the ZWO ASI 120MC-S or the 224MC.  The field of view would be very much smaller and so far more 'zoomed in' but zoom doesn't imply resolution. The 'magnification' (I've read the threads about whether its real magnification or not!) produced by the telescope would, of course, be the same.  However, I assume that resolution is at least partly controlled by pixel density?  The pixel size of the above cameras is 3.75µm against 6.25µm for the Canon.  Would this then give me 1 2/3x the resolution of the Canon?  I'm hoping that an astro camera would work better with a Barlow so could match the 'magnification' but in theory be able to zoom in almost twice as far without losing detail.  What do people think?

Also, do people have views on whether the ASI224 is worth the additional £70 over the 120? The main advantages seem to be fps and noise which may not be that important for lunar (though when I get to planetary I guess would be more useful).

I'd be grateful for thoughts on the above and any suggestions for getting higher resolution.  Major expenditure (eg a 'bigger' telescope)  is not an option.

Many thanks in advance.

[AstroMuni]

Resolution is dependant on the scope in use. But getting more detail can be achieved by taking more images and stacking. Focus also plays a big role.

[vlaiv]

There is a limit imposed by physics of light and size of aperture on level of detail that you can record.

For any given pixel size - there is maximum focal length that you can use on certain aperture.

In most cases, limiting factor is atmosphere and if you want to shoot highest resolution images possible for a given telescope - you need to use lucky imaging approach. Again, in order to do that - you need very specific type of camera - dedicated planetary camera.

For dedicated planetary camera, few things are important:

- QE of sensor

- low read noise

- fast readout rate

Pixel size is not important as you can use barlow lens to get required focal length for any pixel size.

Moon does not have any significant color detail so you can use mono sensors that are more sensitive.

ASI224 is excellent choice because it has great stats on above quoted three properties. It has good QE at around 80%, it has very low read noise and it can deliver extraordinary high FPS.

Say you go for camera like ASI224 and you use 3.75µm pixel size. You then need ~ F/15 telescope or in your case x2.5 barlow. Use exposure length of about 5-6ms and get tens of thousands of frames. Stack the best 10-15% (judge how many based on quality of your frames) and apply some sort of frequency restoration process (say wavelet sharpening).

This will get you very close to theoretical limit of 200mm aperture.

Using that technique, I captured this image:

http://serve.trimacka.net/astro/Forum/2020-07-30/moon.png

(SGL no longer allows embedding non https images - so I'm providing link only - but you can view it in full resolution by enlarging it to 100%).

Above image was taken with 100mm telescope. This means that you should be able to go twice higher resolution with 200mm telescope.

[CraigT82]

Lucky imaging with a planetary camera is definitely the way to go, along with utilising a barlow. There is no point getting closer in by adding more barlow power and capturing still images as the atmosphere will be the limiting factor, you need to capture video and selct/stack the best frames as outlined by Vlaiv in order to get past the atmosphere limit.

Doing it that way you will get much closer in (i.e. capture fewer arc seconds of sky per pixel, due both to more barlow power and smaller pixels) but unfortunately you wont get anywhere near as much of the lunar surface in shot as you do with the 5D.  Saying that there are astro cameras with large APS-C size sensors (eg. ZWO ASI1600) but they do cost a small fortune.

Vlaiv's image linked to above is a cracker, especially for a 100mm scope.  I've taken a screenshot of his image alongside my own image captured with a 220mm scope to give you a taste of the resolving power increase going from 100mm to 200mm (ish). Do bear in mind that at this image scale I had to shoot 25 panels and stitch them together to cover just half of the lunar surface (full image can be seen my my astrobin link in my signature). As I mentioned you could get larger sensor cameras but costs go up rapidly.

 

 

[vlaiv]

16 minutes ago, CraigT82 said:

  I've taken a screenshot of his image alongside my own image captured with a 220mm scope to give you a taste of the resolving power increase going from 100mm to 200mm (ish).

Your image really is a beauty. I just love that level of sharpness and detail in an image.

[andyrawlins]

4 hours ago, vlaiv said:

There is a limit imposed by physics of light and size of aperture on level of detail that you can record.

For any given pixel size - there is maximum focal length that you can use on certain aperture.

In most cases, limiting factor is atmosphere and if you want to shoot highest resolution images possible for a given telescope - you need to use lucky imaging approach. Again, in order to do that - you need very specific type of camera - dedicated planetary camera.

For dedicated planetary camera, few things are important:

- QE of sensor

- low read noise

- fast readout rate

Pixel size is not important as you can use barlow lens to get required focal length for any pixel size.

Moon does not have any significant color detail so you can use mono sensors that are more sensitive.

ASI224 is excellent choice because it has great stats on above quoted three properties. It has good QE at around 80%, it has very low read noise and it can deliver extraordinary high FPS.

Say you go for camera like ASI224 and you use 3.75µm pixel size. You then need ~ F/15 telescope or in your case x2.5 barlow. Use exposure length of about 5-6ms and get tens of thousands of frames. Stack the best 10-15% (judge how many based on quality of your frames) and apply some sort of frequency restoration process (say wavelet sharpening).

This will get you very close to theoretical limit of 200mm aperture.

Using that technique, I captured this image:

http://serve.trimacka.net/astro/Forum/2020-07-30/moon.png

(SGL no longer allows embedding non https images - so I'm providing link only - but you can view it in full resolution by enlarging it to 100%).

Above image was taken with 100mm telescope. This means that you should be able to go twice higher resolution with 200mm telescope.

Brilliant, thanks vlaiv.  I'd be very happy with a picture like that  Was it taken with a 224? Is it a mosaic? I thought they have a very tight fov. 

[andyrawlins]

3 hours ago, CraigT82 said:

Lucky imaging with a planetary camera is definitely the way to go, along with utilising a barlow. There is no point getting closer in by adding more barlow power and capturing still images as the atmosphere will be the limiting factor, you need to capture video and selct/stack the best frames as outlined by Vlaiv in order to get past the atmosphere limit.

Doing it that way you will get much closer in (i.e. capture fewer arc seconds of sky per pixel, due both to more barlow power and smaller pixels) but unfortunately you wont get anywhere near as much of the lunar surface in shot as you do with the 5D.  Saying that there are astro cameras with large APS-C size sensors (eg. ZWO ASI1600) but they do cost a small fortune.

Vlaiv's image linked to above is a cracker, especially for a 100mm scope.  I've taken a screenshot of his image alongside my own image captured with a 220mm scope to give you a taste of the resolving power increase going from 100mm to 200mm (ish). Do bear in mind that at this image scale I had to shoot 25 panels and stitch them together to cover just half of the lunar surface (full image can be seen my my astrobin link in my signature). As I mentioned you could get larger sensor cameras but costs go up rapidly.

 

 

Another excellent shot Craig, thanks very much.  I'm not too bothered about getting large lunarscapes, its detail I'm interested in. 

What tools do you guys use for stacking etc?  I have always used Registax but people now seem to use something else for stacking (eg Autostakkert) and then Registax for wavelets.  Why is that?  I've been dabbling with APT which seems to be an amazing tool.

[vlaiv]

12 minutes ago, andyrawlins said:

Brilliant, thanks vlaiv.  I'd be very happy with a picture like that  Was it taken with a 224? Is it a mosaic? I thought they have a very tight fov. 

No, I used ASI178mc and Skymax102 maksutov telescope.

Yes, it was mosaic, I think at least 6 panels.

I use Autostakkert!3 for stacking - as it is easiest to use and gives best results. Other than that - Pipp (which is planetary imaging pre processor - handy tool).  Registax wavelet sharpening and Gimp for touch-ups

In this particular instance I used Microsoft ICE for stitching panels since I used AltAz mount and there was some field rotation between panels, but often I use ImageJ and its plugins for mosaic stitching.

[CraigT82]

Yes same for me, Autostakkert3 for stacking, registax 6 for wavelets and the gimp for finishing. There is another software I've been using too that is pretty powerful called Astrosurface.

[vlaiv]

23 minutes ago, CraigT82 said:

Yes same for me, Autostakkert3 for stacking, registax 6 for wavelets and the gimp for finishing. There is another software I've been using too that is pretty powerful called Astrosurface.

I did not know about Astrosurface - will check it out.

[andyrawlins]

7 hours ago, vlaiv said:

Say you go for camera like ASI224 and you use 3.75µm pixel size. You then need ~ F/15 telescope or in your case x2.5 barlow.

I've heard about this pixel ratio but I don't fully understand it.  Could you point me to chapter and verse please?  

[AstroMuni]

10 hours ago, andyrawlins said:

I've heard about this pixel ratio but I don't fully understand it.  Could you point me to chapter and verse please?  

Here is a link that gives you info on relationship between Telescope & Camera https://astronomy.tools/calculators/ccd_suitability and this link should help you with the calculations https://astronomy.tools/calculators/ccd

Edited January 28, 2021 by AstroMuni

[vlaiv]

12 hours ago, andyrawlins said:

I've heard about this pixel ratio but I don't fully understand it.  Could you point me to chapter and verse please?  

Not sure which part you don't understand, so I'll do a quick intro on all three possible.

1. Sampling rate

Telescope + camera is projection device and angle in the sky is mapped onto linear distance on the surface of the sensor. Scale factor related to this mapping can be expressed as arc seconds (measure of angle - 1/60th of arc minute which is itself 1/60th of a degree) per pixel.

It can be thought of as "zoom" although technically it is not zoom (zoom is ratio of angles, and here we convert angles to linear distance)

Above is formula to calculate it. Real formula should include Tan trig function but we are using small angle approximation that tan x = x for small angles (like mentioned arc seconds) for simplicity (above formula is good enough).

It is useful both in planetary and regular imaging

2. F/ratio of telescope, sometimes also called "speed" of telescope is ratio of aperture and focal length. It is useful in these calculations because sampling rate depends on focal length and actual detail resolved depends on aperture size. It turns out that certain pixel size has fixed F/ratio for what is called critical sampling

3. Telescope resolution / critical sampling

Maximum details that can be resolved by a telescope depends on aperture size. In fact there is relationship that defines maximum spatial frequency of signal at focal plane and is given as 1/ (lambda * f_ratio) - where lambda is wavelength of light and f_ratio is F/ratio of telescope.

Nyquist-Shannon sampling theorem says that we should be sampling the band limited signal at twice its highest frequency (regardless of what you might read elsewhere - this applies to 2d case as well), we can derive expression for needed F/ratio for critical sampling.

Critical sampling is resolution at which we can capture all theoretically possible detail for telescope of given aperture.

If 1 / (lambda * f_ratio) is highest frequency component, then associated wavelength is lambda * f_ratio. We need to sample twice per that wavelength (two times higher frequency) - so our pixel needs to be half that.

lambda * f_ratio = 2 * pixel_size

f_ratio = 2 * pixel_size / lambda

lambda is usually taken to be 500nm for broadband / RGB type of imaging, although you can put actual wavelength if you are using narrowband filter like Ha to stabilize seeing (in that case lambda would be 656nm).

f_ratio = 2 * 3.75mm / 0.5µm = 15

So F/ratio for critical sampling with  3.75µm pixel size = F/15

[andyrawlins]

10 hours ago, AstroMuni said:

Here is a link that gives you info on relationship between Telescope & Camera https://astronomy.tools/calculators/ccd_suitability and this link should help you with the calculations https://astronomy.tools/calculators/ccd

Brilliant, thanks AstroMuni, I hadn't come across those sites.   Its good to see Nyquist here -  I know him well from my interest in HiFi - though in that sphere many now consider his 2x sampling requirement to be too low for music reproduction. A quite different environment though for a very different forum

[vlaiv]

2 minutes ago, andyrawlins said:

Brilliant, thanks AstroMuni, I hadn't come across those sites.   Its good to see Nyquist here -  I know him well from my interest in HiFi - though in that sphere many now consider his 2x sampling requirement to be too low for music reproduction. A quite different environment though for a very different forum

Yes, except people seem not to understand Nyquist theorem. For example:

is flawed on several levels.

Why would we connect seeing FWHM with Nyquist sampling frequency by factor of x2?

What is "analog signal" and why would we want to image with resolution of 1/3 of that?

Stars are never squares - that is fallacy produced by nearest neighbor sampling when enlarging images. Pixels are point samples and not little squares ...

Etc ...

[andyrawlins]

7 hours ago, vlaiv said:

Not sure which part you don't understand, so I'll do a quick intro on all three possible.

1. Sampling rate

Telescope + camera is projection device and angle in the sky is mapped onto linear distance on the surface of the sensor. Scale factor related to this mapping can be expressed as arc seconds (measure of angle - 1/60th of arc minute which is itself 1/60th of a degree) per pixel.

It can be thought of as "zoom" although technically it is not zoom (zoom is ratio of angles, and here we convert angles to linear distance)

Above is formula to calculate it. Real formula should include Tan trig function but we are using small angle approximation that tan x = x for small angles (like mentioned arc seconds) for simplicity (above formula is good enough).

It is useful both in planetary and regular imaging

2. F/ratio of telescope, sometimes also called "speed" of telescope is ratio of aperture and focal length. It is useful in these calculations because sampling rate depends on focal length and actual detail resolved depends on aperture size. It turns out that certain pixel size has fixed F/ratio for what is called critical sampling

3. Telescope resolution / critical sampling

Maximum details that can be resolved by a telescope depends on aperture size. In fact there is relationship that defines maximum spatial frequency of signal at focal plane and is given as 1/ (lambda * f_ratio) - where lambda is wavelength of light and f_ratio is F/ratio of telescope.

Nyquist-Shannon sampling theorem says that we should be sampling the band limited signal at twice its highest frequency (regardless of what you might read elsewhere - this applies to 2d case as well), we can derive expression for needed F/ratio for critical sampling.

Critical sampling is resolution at which we can capture all theoretically possible detail for telescope of given aperture.

If 1 / (lambda * f_ratio) is highest frequency component, then associated wavelength is lambda * f_ratio. We need to sample twice per that wavelength (two times higher frequency) - so our pixel needs to be half that.

lambda * f_ratio = 2 * pixel_size

f_ratio = 2 * pixel_size / lambda

lambda is usually taken to be 500nm for broadband / RGB type of imaging, although you can put actual wavelength if you are using narrowband filter like Ha to stabilize seeing (in that case lambda would be 656nm).

f_ratio = 2 * 3.75mm / 0.5µm = 15

So F/ratio for critical sampling with  3.75µm pixel size = F/15

Brilliant vlaiv, a very full description.  The Astronomy Tools calculator that AstroMuni pointed me to suggests that if anything I should use a focus reducer in this set up rather than a Barlow.  Is this related to the default of 'ok seeing conditions'.  if I change it to 'exceptional conditions' it does indeed suggest a 2.5x Barlow.  

[CraigT82]

8 minutes ago, andyrawlins said:

Brilliant vlaiv, a very full description.  The Astronomy Tools calculator that AstroMuni pointed me to suggests that if anything I should use a focus reducer in this set up rather than a Barlow.  Is this related to the default of 'ok seeing conditions'.  if I change it to 'exceptional conditions' it does indeed suggest a 2.5x Barlow.  

That calculator is built for still imaging (i.e. long exposure, DSO). Different set of rules for lucky imaging. Xposures are so short you don't even need a tracking mount! 

[vlaiv]

12 minutes ago, andyrawlins said:

Brilliant vlaiv, a very full description.  The Astronomy Tools calculator that AstroMuni pointed me to suggests that if anything I should use a focus reducer in this set up rather than a Barlow.  Is this related to the default of 'ok seeing conditions'.  if I change it to 'exceptional conditions' it does indeed suggest a 2.5x Barlow.  

That calculator is trying to give you best sampling rate for long exposure astrophotography - much different topic than planetary.

We can discuss that topic as well and how to use Nyquist sampling theorem in that context if you wish - but that is not something that you'll need to get best lunar shots.

[jetstream]

8 hours ago, vlaiv said:

Critical sampling is resolution at which we can capture all theoretically possible detail for telescope of given aperture.

Ok, in the example formula, f_ratio = 2 * 3.75mm / 0.5µm = 15  where did the .5um come from?(lambda??)

Just trying to understand so I can do it myself.

[CraigT82]

8 minutes ago, jetstream said:

Ok, in the example formula, f_ratio = 2 * 3.75mm / 0.5µm = 15  where did the .5um come from?(lambda??)

Just trying to understand so I can do it myself.

0.5 microns is 500 nanometers. Wavelength of green light (lambda is Greek symbol for wavelength) 

[jetstream]

19 minutes ago, CraigT82 said:

0.5 microns is 500 nanometers. Wavelength of green light (lambda is Greek symbol for wavelength) 

Ok I knew that

Thanks Craig.

[jetstream]

@CraigT82 do we adjust any of this for seeing conditions?

[vlaiv]

18 minutes ago, jetstream said:

@CraigT82 do we adjust any of this for seeing conditions?

Yep, lambda is wavelength of light and I used 500nm or 0.5µm.

If you image with narrowband filter like Ha - you can use that wavelength.

In theory we should do it for 400nm as this wavelength of visible spectrum represents truly highest possible frequency, but in reality - blue part of spectrum is the most affected by atmosphere and going for 400nm would make us oversample quite a bit in red part of spectrum.

There is almost 2:1 relation between highest and lowest frequency of visible part of spectrum (or shortest and longest wavelength) 700:400 = 1.75 : 1 - almost 2:1 - which in turn means that F/ratio for red part is significantly lower than optimum F/ratio for blue part of spectrum. Going for 500nm is just a good compromise for full spectrum planetary critical sampling.

This is completely without influence of atmosphere. Atmosphere only lowers available detail, but we are doing lucky imaging and hope that we will minimize atmospheric impact. As such we need to image so as to capture any possible detail that aperture provides. Whether we will manage to do that on particular occasion depends on luck part of lucky imaging approach

 

[jetstream]

5 minutes ago, vlaiv said:

lucky imaging approach

I like this term 😀

So if I took captures with a color camera and picked an f ratio in reality only one colour would be proper out of this set of frame captures? How much affect does this have I wonder.

Avani explained to me the benefit of the IR filter vs seeing a while back.

[pete_l]

You can use eyepiece projection to get a larger apparent  focal length beyond what prime focus photography will give you.

If you are using an unmodified DSLR, it probably has a moire filter in front of the sensor. This acts to blur images thus reducing the resolution of the camera below what you would expect from doing the maths on pixel size / focal length.

However, you ultimately run up against the effects of the atmosphere and then against diffraction of the light coming in.  You might get some further improvement by imaging in the infra red, although that will worsen the diffraction effects.