AstroGuider 4.x

[iMustBcrazy]

Excellent! I'll be trying it again this Friday, with a little luck!

[Peter Polakovic]

This is great! And I like Johan's idea -- why determine focus on 1 star when you can average over several? Could that avoid the need to stack exposures during autofocus, making the process 3x quicker?

We tried this approach in a very first AstroGuider, but it was not much better than single star DONUTS turned out to be much better (guiding with all stars in the frame).

Also this may help with another interesting idea. Instead of determining focus by FWHM, N.I.N.A. has a mode where focus can be determined by counting stars. More stars = better focus. I've never tried that (allergy to Windows) but I've been intrigued.

This may be a good idea, I'll try it BTW, we use combination of FWHM, HFD and peak value for focusim. Both FWHM and HFD are a rough approximation (otherwise it is too slow), so it help a little bit to combine them.

Peter

[JohanB]

Hi Peter,

We tried this approach in a very first AstroGuider, but it was not much better than single star DONUTS turned out to be much better (guiding with all stars in the frame).

Please, do not stop improving on the star selection algorithm, because Donuts is available. The Donuts algorithm has been an issue for me most of the times. The donuts algorithm still produces the occasional unexplained 2px bump every now and then (unless the problem was solved over the summer.

The bump is a real issue, because the 2px deviation spoils the image. Even though it fully corrects in the next exposure, the long exposure time used for Donuts means the image is 2px off for many seconds, leaving visible trails.

Regards, Johan

[JohanB]

Checked my old e-mails; actually it is a 1px bump

[Peter Polakovic]

Don't worry, we'll not drop it Nevertheless, Rumen fixed also some issues in DONUTS, maybe it will work better now. Peter

[JohanB]

Hi Peter,

Great! Will give it a shot next time. Would love to be able to use Donuts, since it does seem to provide much higher quality.

Regards, Johan

[Peter Polakovic]

INDIGO updated, star detection improved:

http://download.cloudmakers.eu/AstroGuider_4.0_541.dmg

[Peter Polakovic]

Along with adding hot pixels/hot columns to CCD simulator we found, that DONUTS algorithm doesn't work correctly if there is hot pixel or hot column in the frame. Make sure to avoid them by selecting clean subframe. We're working on the solution... Peter

[Peter Polakovic]

In a next few days AG4 will identify that the guiding is settled down after dithering automatically. No need to set any timelapse on AI4 side... Coming soon

Screenshot 2020-09-20 at 16.30.10.jpg (137.88 KiB) Viewed 131 times

[aaron]

Oh man that will be fantastic! Very smart!

Harder to calculate estimated competition time perhaps... can the guider agent report “last settling time” in the INDIGO api? That would help for my app, which is getting a lot closer.