SharpCap Forums

Same technique as previous post viewtopic.php?f=16&t=4136 different only in minor details but this time showing the 'speckle' images of the core of each of three globular clusters.

M92 and M3 have quite compact cores where the additional resolution of lucky imaging has uncovered quite a lot of additional detail and has added punch to the overall high dynamic range compositions. M13 has a more open core and the 0.13s exposures hasn't resolved much that isn't anyway resolved in the stack of 10s images. Overall the technique seems to work OK improving resolution from an FWHM of about 2.5-3 arc sec down to about 1.5 arc sec. At an exposure time of 130 ms about 5-7% of the exposures were selected as 'good'.

For the overall compositions the tricky bit is to get the masking right so as to get the small 'lucky' images seamlessly incorporated. It would have been better no doubt to have collected more data and to have stacked more frames since the SNR is not that high at only 20-25s or so of total exposure.

It would be interesting to see just how far 'deep sky' lucky imaging could be taken. With say a large dobsonian - 14-20 inch, a sensitive CMOS camera and at F 4.0 or less I suspect that it would resolve better than conventional imaging for the brighter planetary nebula and say the core of M42.

TimH

Very cool again Tim!!
From the little experimenting I did with lucky imaging, I do know that the amount of captures indeed is key. For the same object (an open star cluster) is did a set of 500 captures and a set of 5000. The 5000 one was SO much better.
Took of course way longer to process with stacking but it was worth it.

Menno

Nice work Tim.

It would be interesting to see just how far 'deep sky' lucky imaging could be taken. With say a large dobsonian - 14-20 inch, a sensitive CMOS camera and at F 4.0 or less

Have a look at https://www.cloudynights.com/topic/5338 ... second-16/ see post #20. Emil author of Autostakkert, user = MvZ on Cloudy Nights. He is producing images like this https://www.astrokraai.nl/viewimages.ph ... 6&s_fromcn with a 400mm dob.

Dave

Menno555 wrote: Sun May 23, 2021 7:19 am i did a set of 500 captures and a set of 5000. The 5000 one was SO much better.

Thank you Menno So far have not gone beyond about 8 minutes of video capture for a file size of about 6 Gb (about 3000 captures). After selecting the 'lucky' ~ 5% (~ 150) the final total exposure of stacked images then comes down to less than 20s. As you say I am sure things could be much improved - and the final image made at least 3X less noisy - with something like a ten fold upscale. The file sizes and handling will get more difficult though. Further critical questions that I would like to better understand - "How to accurately and efficiently select the best images from many thousands ? " and "What is the relationship between the shortness of the exposure and the percentage of frames that can be selected as 'good' ?"

turfpit wrote: Sun May 23, 2021 3:58 pm Have a look at https://www.cloudynights.com/topic/5338 ... second-16/ see post #20. Emil author of Autostakkert, user = MvZ on Cloudy Nights. He is producing images like this https://www.astrokraai.nl/viewimages.ph ... 6&s_fromcn with a 400mm dob.

Thank you Dave. That is a really pertinent thread that you highlight! An amazingly sharp image of M64 using 1s exposures. It is interesting that it seems to have worked quite so well with exposures that long - i.e. I thought that the dogma was that lucky imaging starts at about 100ms and less - maybe also very good seeing when the M64 image was taken?

Also wonder whether the whole lucky imaging approach should anyway be thought of more as a continuum i.e. it might also 'work' at longer times up to a second or so but just less frequently and down to a less good final resolution? For example- whereas with < 10ms frames (e.g your moon work) one might reasonably expect to select 40% of the frames to resolve right down to the Dawes limit, with 100 ms frames one might find that only 5% can be selected? Maybe 1 and 2 second frames still offer some sharpness benefit (over conventional longer frames) but one that might take more severe pruning to realise and only to a somewhat lesser level of improvement?

I am particularly interested in these questions because my imaging started out with an old 10 inch Dobsonian. Necessarily using short 3s exposures I often seemed to get images of the brighter objects that were satisfyingly sharp - but that obviously lacked depth into the fainter regions. It would be good to get the best of both worlds on more objects by using 'lucky image' luminance to add into and inform the detail within longer and deeper images.

TimH