I tried some new settings a couple nights ago (when I had a break in the clouds) that I was quite happy with. The first thing I corrected from previous sessions was to make the initial step larger than before. I have a Moonlite, and have found setting the step to 10 makes sure that it jumps far enough away from focus that it can truly find the sharpest point. When I had it set to 5 or even smaller, it would not move the focus far enough to find true focus at times. It would center around a false "tight" anomaly such as the seeing being better for that frame or two, and then not move far enough away from that to notice that is was getting better or worse. It is kind of hard to explain, but imagine this, if you have a single star and you take an image of it and get a half value of 2, then you step 2 steps closer focus and get a 2.05, so it moves two steps out and gets a 2.10, it can end up centering on 2, but if it moved 10 steps in first, it has a better chance of seeing actual change outside of atmospheric conditions, and might get a 1.95, which would then cause it to keep searching (correctly). Nothing magic about 10, it just worked well for me most of 2 different nights, so I am leaving it there now.
The next thing I changed was to do the full frame circle with the deleted circle center. I don't recall the percentage numbers, around 50% for the outside circle, and the inside one big enough to block the area around the nebula. The algorithm then selected about 8-10 stars and proceeded to focus. This seems to work really well as it uses the full area around the object to obtain focus without accidentally focusing on part of the object itself. I also used this method for 2 nights, and think I have found my new default settings. A caveat, doing the capture of that much data to focus with is a good deal slower than framing a single star, but I found it to be more accurate as well. I also do 2 frames per adjustment instead of the default 1.
First, I really like that the object name populates the prefix field. It saves a lot of typing. I discovered tonight however, that on a comet hunt, the names with a "/" in them should be changed to a - or _ to prevent file system directory naming issues on *nix type systems
Without this, it creates a directory 38P and continues to overwrite the same file without increasing the number. I am fairly confident it is because when it goes to look for the next file name, it doesn't find the current one in the structure that is created, but it would be better just to modify the char from "/",
such as in the attachment.
Running 3.0.0 Dec 21 build, working great so far tonight.
I guide with a ZWO 290 mini. Works great, every time. I also have a 290 color usb-3 that works as a guide, every time as well.