Wednesday night was the first night with any clear weather since 2.5.1.3 was released and I had a short window of about an hour early on to test the new auto-focus. I think my results should not be considered as very reliable since 1) clouds kept drifting by, and 2) the collimation on my RC12 is really poor. Anyway, here they are:
Focus routines seemed to be finding stars very reliably and consistently, on both the RC12 and my SV102 mm refractor.
Results were very comparable to earlier releases. I set the minimum star size to 4. Clouds did not allow me time to try other settings. I did 5-6 runs on each scope. The refractor V curves were excellent. The RC12 curves were some good, some fair, very much like versions going back several months.
I did not try a focus range large enough on the RC12 to produce donuts because of the poor collimation.
I can post the focus packs if you want them, but I don’t think they will be of much use.
same here - i actually did post AFPacks in the thread about 2.5.1.1 focus not working but as you indicate they are all pretty uneventful except for the first one. my refractor is in storage so it’s good to hear that the new routines are working well there as well.
edit: your link to the AFPacks seems to be missing…?
Hi Rob,
I think your edit about the missing link to AFPacks is to me. If so, I said “I can post the focus packs if you want them” so there is no link.
I don’t think they will help Ken. I would prefer to send ones I get when seeing is good and I have a good collimation, which I will be doing next week.
Jerry
to be honest they could be useful as they are probably “difficult” AF cases. a recent change to the star detection algorithm was made to detect less circular stars…
Less circular star detection may indeed be part of the enhancement, but I think the major change was to try to detect donut stars when central obstruction scopes are way out of focus. My runs did not go that far out of focus, so are not able to help with that testing.
I did some more experiments and had a chance to use 2.5.1.3 using the settings I had arrived at with 2.5.1.2. It proved one thing; developing an AF routine that works in all cases is very tricky. Between these and the prior tests I did a full set of dark frames, in 2-second intervals and integrated 10 frames for each exposure.
With short exposures on a new target, even though I could see the stars plainly, was a complete wipe out, with no detected stars at any scale. I flipped back to 2.5.1.2 but it made no difference. I increased my exposure from 2 to 8 seconds and I started to have success. I’m guessing that the star intensities for this target were less than those earlier in the week and that the AF routine sets a minimum star peak threshold in its detection algorithm. Seeing was good and I did get a few good V-curves in the end with reasonable consistency. In the cases where no stars were detected, the auto screen stretch showed good round stars, covering a 10 or so pixels.
If my assumption is correct, it would be useful to know what the min peak intensity threshold is. Since the AF exposure is not adaptable (as Focusmax can be), if we know what the min intensity is, one can do a quick acquisition, note the peak 21x21 intensity and pitch the AF exposure to it.
As it currently stands, the AF exposures for each filter are fixed in the equipment profile, with the AF points. I wonder if it is worth considering ratiometric exposures i.e. if the standard exposures are 2,4 and 8 seconds for L, RGB and NB filters and you manually set the AF exposure at 3 seconds for L, the other filter exposure times are changed to 6 and 12 seconds?
