There's a
new thread over on cloudy nights
on how a user wrote their own capture and guiding program that does all guiding through the primary imager with no guide scope or OAG/ONAG.
It's extremely clever and seems like the future of CMOS capture should go down this route.
It works by capturing a frame every 5 seconds, sends a guide command, measures FWHM, and chooses to keep or discard the image based on quality of the FWHM, after it has captured 10 minutes worth of 5 second exposures that meet the quality requirements, it automatically creates a stacked image, for a single 10 minute exposure.
This type of system simplifies setup and capturing, and ensures a high degree of accuracy only found with very high end imaging systems. Would be interesting to see a modified version of EKOS that operates like this.
The code was added by Jasem and Csaba last year (Spring 2017), but was not finished yet. There is too much on the to do list. . . I know I have been more than busy with lots of other things and so have they. Want to volunteer?
Anyway, my understanding from Bob (the developer) was that the plan was/is to use this code with a guide camera to get more accurate guiding from the guide camera. Thats also how it was/is integrated into Kstars in the guide module. I didn't think that he planned to use it with the main imager. In most cases with astro cameras (with a few notable exceptions like ZWO maybe) it is better to take sub exposures that are longer than the amount of time you would want for a guide image
The principle being applied in OP's post is a form of lucky imaging. I have been told it can produce impressive results, but is a great deal of data to handle.
Lucky imaging also requires a decent SNR to start with. Else whatever number of frames you will capture won't be enough to detach the object from the background, even if the theory is backed by statistics. This said, this method is certainly very effective for tight narrowband imagers.