"Lucky" imaging for galaxies

Tony Gondola · Jun 10, 2026, 03:20 AM

Although I wouldn’t call it true lucky imaging it is an approach if you want as much detail as possible in your images. First you have to have the right sampling. I would base that more on your hardware and optics rather than what your “average seeing” is supposed to be. For your setup I would not use the reducer but rather shoot at prime. That will put you right were you want to be for 3.7 micron pixels. Exposure times will range from a few seconds to as much as 15 or 30 sec. You’ll also want to pump up the gain to get the read noise down as low as possible. With a 533 that will be around 200 to 300 but keep an eye on your star cores as the trade off will be lower dynamic range.

With such short exposure times you are going to gather a lot of frames but that’s what you want. Also, you really want to look for those nights where the seeing is at least good for your area. Using a weather app like the clear sky chart or atmospheric will give you that data. You want to keep an eye on the position of the jet stream too Also avoid shooting at low elevations below 60 degrees. That means for an object transiting directly overhead you’ll get about 4 hours worth of data per night. Of course your focus and guiding will need to be spot on.

Once you have your data you’ll want to deeply cull your subs. There’s a lot of different ways to do that depending on your software. You want to cull based on FWHM. You might have average seeing for 2” but if you’re done everything else right there will be a number that were taken during those short spells when it calms down. It might take several nights to get enough good frames for a decent S/N. You can get improved results with this technique but don’t expect miracles as this isn’t really lucky imaging. Pick bright objects and get a lot of high quality data.

An easier road to improved detail would be to shoot at F/10 with normal sub exposure length, whatever that is for your skies. The trick is to stack two images from the same original data set. One including all usable frames for S/N and one deeply culled for the best resolution. You can then take the high frequency information from the deeply culled image and combine that with you high S/N image. Techniques for doing that will depend on your software.

All this is fun to play with and is worth doing, even if it’s just to find the limits for your hardware and location.

Good ideas all around, appreciate it! I’ve been very hesitant to shoot at f/10 due to my guiding numbers. I think the best I can do right now is around .75 to .85” with my eq-g, with occasional spikes up around .9 to 1”. I have had longer periods of around .6” guiding so i’m hopeful I can get it refined a bit more. I bought this mount used, and I love it, but I don’t know how much to expect from it. I’ve looked into getting belt mod kits but i’m a little worried it could make things worse.

All this said, I’m going to give it a shot, why not! I really like the idea of getting a high s/n image and a high frequency detail image and compositing them together. I had played around with using high pass filters and basically doing that with my current images, but the data is just not there.

Tony Gondola · Jun 10, 2026, 01:31 PM

Yes, you will have to get the guiding numbers down a bit. The short exposures will help with that but with better guiding you’ll have to cull less. If you can tell me what your guiding setup is I might have some ways to improve it. I also forgot to mention that process sharpening is an important part of the process. Cosmic Clarity or BlurX is a very important part of the workflow.

Thanks, Tony. Yeah I downloaded the trial version of Pixinsight and BlurXTerminator and just starting to play with it.

For my guiding setup, I have a 240mm svbony guide scope and a sv305c guide camera. I realize I should move to an oag and mono guide cam. I’m using phd2 for guiding, multi star and generally just default settings from phd2 and assigning any values suggested from guiding assistant, except dec backlash suggestion since that did not help in the past.

I’m looking to try predictive pec some more as well, either within phd2 or through eqmod. My backlash is high in the dec axis. Guiding assistant says around 1700ms and that was after adjusting the grub screws to reduce some minimal play that I could feel while rocking my ota back and forth. Before the adjustment, it was 2700. I also wonder if some of that backlash is caused by stiction. I think my mount has the black grease i’ve heard others talk about. It’s hard for me to say what normal or optimal is for these mounts since I have no frame of reference, but while the motion of both the ra and dec axis is smooth, it does take an initial nudge to get the mount moving with the clutches off. It makes balancing the scope a bit tricky. What I ended up doing was finding the extremes at both ends of my counter weight positions where either the weight of the ota or counter weights can offset the balance without a nudge from me. Then I just add a slight bias depending on where i’m shooting.

Tony Gondola · Jun 10, 2026, 04:15 PM

Going to an OAG would probably improve your guiding numbers. I would not use the mount’s PEC to improve guiding because it can often just end up fighting PD2. You can use the PEC option that’s available within PHD2 though. That might be of some benefit.

Yeah i’ll need to just suck it up and get an oag. I’ll give the phd2 pec a shot as well. I’ve heard that about the eqmod pec as well, there’s some conflicting opinions on that. Sure would be nice to use the pec curve without training phd2 every session but I get the feeling you're right and it will just fight phd2. I wish there was a way to test these things in the daytime. You only have so much time at night.

Thanks again!

I would prefer phd2 predictive algorithm over the mount pec for at least one reason : as phd processes position value continuously, the estimates of PE parameters are adjusted in real time and are probably more precise.