New attempt :)

[NickK]

Ok, so from the same session Lunt60 Ha with ATIK Titan, my OSX drivers, Registax aligning (and drizzle) using Wine then stacking single aligned pngs using PI.

I think that PI provides better control but the drizzle from registax is not great!

[NickK]

Attempting more processing...

.. now i want my own lunt lol...

[JamesF]

I think you're probably oversampling there anyhow, so I'm not convinced drizzle buys you much. I need to look into that a bit more.

James

[allcart]

Can you explain over sampling James. And do you know what causes the lines and artefacts around the edges.?

[Montana]

Excellent proms

The lines round the edge are just part of the stacking program, the more movement, the more you get. I just cut mine off to tidy the picture.

Alexandra

[JamesF]

Can you explain over sampling James. And do you know what causes the lines and artefacts around the edges.?

A telescope has a maximum resolution that is dependent on its aperture and the frequency of the light you're "sampling" to create an image. That effectively means that at some point it doesn't matter how much bigger you make the image, you can't get any more detail because you've reached the limit resolution of the telescope and you're seeing all the detail there is to see.

Obviously a camera has a maximum resolution too -- the size of one pixel. It can't capture an image of anything smaller than that. If the size of the camera pixels matches the size at the image plane of the smallest thing the telescope can resolve then you've reached a balance point between the capabilities of the telescope and camera. If your camera pixels are any bigger you'll be "undersampling" because you're not capturing detail that's there to be seen. If your camera pixels are any smaller then you're "oversampling" because you'll actually get an image of the smallest things the telescope can resolve spread across more than one pixel.

In some respects oversampling might be considered wasteful because you can get exactly the same effect by scaling the image in processing, but there's a thing called Nyquist's Sampling Theorem that says you ideally need to sample at twice the resolution to get good reproduction and therefore a bit of oversampling is actually a good thing to do.

The formulae for resolution are actually relatively straightforward and it's not too much painful maths to work out where the balance point is -- in fact given the pixel size of the camera it's quite simple to work out the focal ratio where this point occurs. If the focal ratio gets slower then you're oversampling and faster means you're undersampling. That's the theory, anyhow. Of course there's another little wrinkle here in that the seeing also acts to limit resolution. I do tend to ignore that and it might be interesting to see what effect it might have on my imaging if I took it into account. It's easier to do the maths than to try to estimate how good the seeing is though

I recently read a paper on drizzle by someone who ought to know what they were on about suggesting that it was only really effective if the initial image was undersampled. If the image is oversampled then there may be no data for the drizzle process to work with other than what you already see. My personal experience with planetary imaging backs this up. I try to oversample by at least a factor of 2 and when I've tried drizzle I seem to get a bigger image but no more detail. I know Darryl, Stuart and a few other planetary imagers posting here use it though and they are certainly oversampling too, so it would be interesting to find out a bit more about what's going on there.

James

[JamesF]

I've just looked up all the figures since I have some time whilst a large amount of data uploads to some servers

I have an aperture of 60mm and focal length of 500mm for the LS60 and a pixel size of 7.4um for the Titan camera. The Ha emission line is 656nm.

That gives a focal ratio of just over f/8.3 for the LS60, and the resolution of the scope and camera should balance out at just under f/11.3. Those figures suggest you are undersampling, but not massively. Drizzle might therefore gain you a little more detail but I'd not expect it to change the world.

I wonder if perhaps the real gains with drizzle are to be had in wide field imaging where the desirable focal ratio might be f/16 but the actual ratio might be f/4. Clearly there's massive undersampling going on there.

If I've made any errors hopefully someone will be along to correct me shortly.

James

[NickK]

Yep, I've been re-reading your post james. the main limitation of 60mm is a little bit of a pain coming from the 105mm aperature.

Dawes limit breaks progressively so for example the pentax at 105mm aperture has a dawes limit of 1.10 arcseconds but I run the scope at 0.88 arcseconds/pixel at 1340mm - the effect of dropping below the limit is blurring once you start sharpening the result can be had.

I have a feeling that performing alignment at this resolution is still accurate - and even though you may have detail spread like a bell curve distribution over the pixels - the detail is recoverable by using the fact that drizzle relies on the misalignment of pixels between frames for the same target to recover information (detail) when stacked on a a higher resolution.

Personally I think registax's drizzle is dire to be honest but it seems the only application that can do FFT-based alignment of images... quite why PI haven't implemented this, in favour of manual two frame alignment is beyond me

[thenews24]

I have pretty much abandoned Registax for solar and lunar surface imaging and use AS!2. For prominence imaging I will use Regiatsax, but whichever program I use I have never used drizzle.

[allcart]

Acouple of very helpful posts there James. Whaat is the maths equation for working this out. I will have the LS60 in a few weeks and will be pairing it up with a DMK41 AU02.

[NickK]

Dawes limit (and links for the equations) can be found here:

(simple using just the aperture without wavelength: http://en.wikipedia....awes'_limit

a more detailed explaination here: http://en.wikipedia....ion#Explanation

This m57 was at 0.88 instead of above the pentax's 1.10 limit can be found here and here

[JamesF]

Acouple of very helpful posts there James. Whaat is the maths equation for working this out. I will have the LS60 in a few weeks and will be pairing it up with a DMK41 AU02.

If you take Rayleigh's formula for angular resolution and combine it with that for plate scale you basically end up with focal ratio = wavelength / pixel size (where wavelength and pixel size are in the same units) for the "balance point" where the smallest resolvable detail is the width of one pixel. If the focal ratio works out at f/15 I therefore try to work at f/30 to get the 2x oversampling to keep Mr Nyquist happy.

James

[NickK]

Excellent proms

The lines round the edge are just part of the stacking program, the more movement, the more you get. I just cut mine off to tidy the picture.

Alexandra

Yup - the shot was on a non-tracking alt/az mount. The sun would drift across the frame.