Focus failure with 2.5.1.3 beta


#1

OK last night i finally had a focus failure. it occurs about 2:01AM in the log. it appears that the HFR was miscalculated somewhere along the way, leading to a restart. at that point the stars were too big for SGP to detect any real stars; in a couple of frames it seems to want to pick up on a hot pixel or some other noise, despite the min star size being set to 4-5 (hard to tell from the slider). the first pack in the zip is the initial failure and the remaining packs have the really large stars which were not detected.

AFpack:

https://goo.gl/hvF56D

log:

https://goo.gl/hNki6D

thanks,

rob


#2

@pfile

I have made some adjustments to the AF routine that I believe make it much more robust (should fix both the issues you reported)… in some cases, it might take a second longer (or so) than it did in 2.5.1.3, but an extra 9 sec per AF run is probably worthy of good focus. I also believe the changes I have made in 2.5.1.4 will do a pretty good job moving the AF routine to an area that is compatible with a great deal more AF data from different cameras, scopes, lighting conditions, etc…

This is no longer the case… SGPro 2.5.1.4 does a pretty good job even out here in donut land. This is the “new” right side of the curve when you got out this far…

A sample of the “new” detection:

Here is the “new” right side of the curve from the original run that presented the problems that got you into donut land (pack 38):


#3

awesome, thanks. i will try it tonight if the weather holds. seem to be some clouds coming in right now. good night for testing though with the moon out.

rob

edit: don’t see .4 on the DL page yet…


#4

OK, testing this now. so far so good. in fact because the focuser position was messed up from the night before, the first focus image was taken 400 steps away from focus, and it still succeeded. it does think for a while about each focus frame but the results seem very accurate.

thanks for all your work on this!


#5

everything seems to have gone well last night. will keep testing as weather allows.


www.mainsequencesoftware.com