How lucky imaging is applied correctly, Sky&Telescope 07/2025 - Forums

Hi all,

I live on the outskirts of a large city (Hamburg, Germany) and so far I have primarily been able to use lucky imaging techniques. In the July 2025 issue of Sky & Telescope I have published a 6-page article on Lucky imaging, check it out Get Lucky in the Deep Sky,

on my website in the technical section there is also a lot written about it.

cheers
Peter

https://pixlimit.com

Glad to see someone is working in that area. I've done traditional lucky imaging, mostly lunar, and know how effective it can be. With that background I try to at least keep it in mind when doing deep sky work by using short subs and culling as needed. Maybe you could call that semi-lucky imaging.

yes exactly, lucky imaging originally comes from planetary and lunar photography and if the conditions are right, such as a bright deep-sky object, long focal lengths, good sampling, you can create a good basis for high-resolution images with short exposure times of around 1 second.

I've dabbled in this kind of DSO imaging and was pleased with the results I got with my C9.25" at f/10. However I found it difficult to use on anything but the brightest objects or those with enough bright-ish stars in the field for stacking. I'd love to use my AST715MC for this with its tiny pixels (1.45 micron) but I'm not sure the FOV will be enough to catch enough stars for stacking. I see you're using SharpCap for the capture. Are you saving your sequences as .SER files?

https://app.astrobin.com/u/ngc1977?i=wqr7b0#gallery

I've experimented a bit with a QHY5iii715C that has 1.49 micron pixels with interesting results:

https://app.astrobin.com/i/joqhd2

But you're right, you have to fight sometimes to get the data to stack. I'd be curious if Peter has any tips on that. For me, I was usually able to stack in Siril using just the 3 star alignment method or if it's really difficult, just one star.

Most of the images in my Astrobin gallery were captured using the Lucky Imaging technique. My best results have been achieved on planetary nebulae, though many galaxies and reflection nebulae also turn out to be quite photogenic. This technique allows me to image deep-sky objects despite the tracking limitations of the AZ GoTo Dobsonian mount

PS. I found some inspiration on the Pixlimit website.

A big dob like yours is perfect for this kind of thing. Your Blue Snowball is wonderful. I don't see your short exposure limit as a guard rail but rather something that's pushing you into an area that you might not otherwise go.

Peter Bresseler:
Hi all,

I live on the outskirts of a large city (Hamburg, Germany) and so far I have primarily been able to use lucky imaging techniques. In the July 2025 issue of Sky & Telescope I have published a 6-page article on Lucky imaging, check it out Get Lucky in the Deep Sky,

on my website in the technical section there is also a lot written about it.

cheers
Peter

https://pixlimit.com

Peter,
That's fantastic and I'd love to read it. Unfortunately it appears to be behind a pay wall.

John

John Hayes:
Peter,
That's fantastic and I'd love to read it. Unfortunately it appears to be behind a pay wall.

John

Hi John,

Baader is currently building a new remote site there under top conditions, and I'm not sure if it's actually a “paywall”. I can put you in touch if you want

cheers
Peter

Peter, to read the article you have to subscribe to S&T. Yes, that's a paywall.

Tony Gondola:
A big dob like yours is perfect for this kind of thing. Your Blue Snowball is wonderful. I don't see your short exposure limit as a guard rail but rather something that's pushing you into an area that you might not otherwise go.

Thank you! I agree, short exposure times make it possible to resolve fine details that would otherwise be smeared by atmospheric seeing.

Maciej Mindziak:
Tony Gondola:
A big dob like yours is perfect for this kind of thing. Your Blue Snowball is wonderful. I don't see your short exposure limit as a guard rail but rather something that's pushing you into an area that you might not otherwise go.

Thank you! I agree, short exposure times make it possible to resolve fine details that would otherwise be smeared by atmospheric seeing.

With that much light to work with it would be interesting to see if, with the brightest objects, if you could push the exposure time down into the 100ms or less range where the best detail is to be found.

John Hayes:
Peter,
That's fantastic and I'd love to read it. Unfortunately it appears to be behind a pay wall.

John

Ah, sorry, I misunderstood. Yes, a paywall.

P.S. I have the original, of course, but I'm not allowed to pass it on.

Tony Gondola:
With that much light to work with it would be interesting to see if, with the brightest objects, if you could push the exposure time down into the 100ms or less range where the best detail is to be found.

I have had good experiences with 500 ms; below that, you cannot achieve a good SNR.

You can't generalize that. Here is a brief explanation

There is already an optimum, for example, a Celestron C11, f/10, camera with a 3.76 µm sensor

3.76/2800*206 = 0.28"/pixel

With an average seeing of 2" we have significant oversampling and thus a poor SNR.

Half the focal length

3.76/1400*206 = 0.55"/pixel

With an average seeing of 2" we still have oversampling, but a better SNR and headroom in image processing.

According to the Nyquist/(sampling theorem) would be optimal

3.76/x*206 = 1; x= 775 mm focal length; with a seeing of 2 and a sensor with 3.76 µm pixels, the ideal focal length according to Nyquist is 775 mm.

Peter Bresseler · Aug 8, 2025, 07:22 PM
3.76/x*206 = 1; x= 775 mm focal length; with a seeing of 2 and a sensor with 3.76 µm pixels, the ideal focal length according to Nyquist is 775 mm.

Hello Peter, where in the equation is the seeing?

What if my seeing is 1’’?

CS, John

Hi John,

What if my seeing is 1’’?

half, in theory,

If you apply the Nyquest scanning theorem to 1“, then you would have to scan with 0.5”. In practice, however, the situation is somewhat different, where an oversampling factor of 3-4 makes sense in order to be able to successfully apply image processing functions (deconvolution, BTX).

Peter Bresseler:
You can't generalize that. Here is a brief explanation

There is already an optimum, for example, a Celestron C11, f/10, camera with a 3.76 µm sensor

3.76/2800*206 = 0.28"/pixel

With an average seeing of 2" we have significant oversampling and thus a poor SNR.

Half the focal length

3.76/1400*206 = 0.55"/pixel

With an average seeing of 2" we still have oversampling, but a better SNR and headroom in image processing.

According to the Nyquist/(sampling theorem) would be optimal

3.76/x*206 = 1; x= 775 mm focal length; with a seeing of 2 and a sensor with 3.76 µm pixels, the ideal focal length according to Nyquist is 775 mm.

Sorry if it's a stupid question, but why would average seeing be relevant to lucky imaging? Other than the amount of frames you're likely to scrap.

Peter Bresseler · Aug 8, 2025 at 08:12 PM
Hi John,
What if my seeing is 1’’?

half, in theory,

If you apply the Nyquest scanning theorem to 1“, then you would have to scan with 0.5”. In practice, however, the situation is somewhat different, where an oversampling factor of 3-4 makes sense in order to be able to successfully apply image processing functions (deconvolution, BTX).

Thanks, Peter. In fact if there is good weather I have often seeing between 1’’-1.2’’. I plan to try lucky imaging with my 6’’ refractor. CS, John

Médéric Hébert:
Peter Bresseler:
You can't generalize that. Here is a brief explanation

There is already an optimum, for example, a Celestron C11, f/10, camera with a 3.76 µm sensor

3.76/2800*206 = 0.28"/pixel

With an average seeing of 2" we have significant oversampling and thus a poor SNR.

Half the focal length

3.76/1400*206 = 0.55"/pixel

With an average seeing of 2" we still have oversampling, but a better SNR and headroom in image processing.

According to the Nyquist/(sampling theorem) would be optimal

3.76/x*206 = 1; x= 775 mm focal length; with a seeing of 2 and a sensor with 3.76 µm pixels, the ideal focal length according to Nyquist is 775 mm.

Sorry if it's a stupid question, but why would average seeing be relevant to lucky imaging? Other than the amount of frames you're likely to scrap.

I think that's an excellent point...

Médéric Hébert:
Peter Bresseler:
You can't generalize that. Here is a brief explanation

There is already an optimum, for example, a Celestron C11, f/10, camera with a 3.76 µm sensor

3.76/2800*206 = 0.28"/pixel

With an average seeing of 2" we have significant oversampling and thus a poor SNR.

Half the focal length

3.76/1400*206 = 0.55"/pixel

With an average seeing of 2" we still have oversampling, but a better SNR and headroom in image processing.

According to the Nyquist/(sampling theorem) would be optimal

3.76/x*206 = 1; x= 775 mm focal length; with a seeing of 2 and a sensor with 3.76 µm pixels, the ideal focal length according to Nyquist is 775 mm.

Sorry if it's a stupid question, but why would average seeing be relevant to lucky imaging? Other than the amount of frames you're likely to scrap.

I think that's an excellent point...

It affects the rate at which images are stacked.

because average seeing is relevant, it has to be taken into account. Seeing 1-2 is rather rare, 2-3 is common (in my experience).

With good seeing, the rate of images that are stacked is higher (>50), with poor seeing it is lower (<20). But then there is a problem with the SNR.

Tony Gondola:
Tony Gondola (Gondola)

Yes, that's true with any application of the Lucky Imaging technique. However, I do think you have to go beyond Nyquist. I would consider your example of a 775mm system with 3.76 pixels as being useless for lucky imaging if your goal is to achieve higher resolution which to me, is the whole point of lucky imaging in the first place. If you are going to take a group of 4 pixels to be the minimum needed to define a star, which is far from showing it as round, the best you will ever do with that system is to resolve blocky details down to 2" if everything else is perfect. A high SNR is great but if you start to push the limits by applying the concepts of Lucky Imaging, at least in part, to deep sky objects. I think are going to have to push the accepted limits a bit. Just as the lunar planetary guys do, you are going to have to lean heavily on the very powerful software tools that are at our disposal. The whole point is to go sharp, not deep. You also have to accept that your sub-exposure rejection rate is going to be high, just as it is in Lunar and Planetary work.

At least with present technology I don't think that deep sky imagers can go as far as the L&P guys do but smart application of some of the precepts can help push the goal of achieving more detailed images.