There is a note in the latest ASIAIR Beta that mentions "optimizing" the auto focus algorithm, so I decided to take the scope out to test it, in spite of terrible smoke in the area (Portland's PM2.5 count is 106 at the moment, and my indoors PM2.5 count is 25 even though there are multiple Dyson air purifiers in the house; pretty much dangerous to be outside without a mask -- I used an R95 mask :-).
It looks like ASIAIR may have stopped using the bogus second pass star sizes to focus, and the autofocus results I get is much more consistent now.
I am testing with a refractor with 60mm aperture and 300 mm focal length, and an ASI2600MC with a pretty generic light pollution filter. Gain of the camera was set to 100.
The EAF step count after "auto focus" resulted in EAF steps of: 30095, 30094, 30096, and 30097. The first two used exposure time of 2 seconds, and the last two used exposure time of 1 second. The last one had tree branches obscuring 1/4 of the frame. The first three were in the Vega region, and the last one in the Sharpless catalog 155 region.
Air tempertaure changed from about 16.1ºC to 15.8ºC. The CFZ is somewhere around 30 to 40 EAF steps, and I had told ASIAIR to use a step size of 10. Except for the smoke, the skies are clear, with darkness hours in between nautical twilight and astronomical twilight (i.e., not completely dark yet).
In one case, I also measured the star size using the Detect star tools for the given EAF position, and one that is 5 EAF steps lower, and the latter gave a star size that is larger than the optimal EAF position.
So, things definitely looking much better than before.
They may have finally abandoned that second pass measurements. This is a path from one of them:
In the past, based on the progression of the measurements in the second pass, the ASIAIR would have used the EAF value from the second green dot from the right (since the third green spot had a larger measured star size), and result in an EAF position that is way early.
Another EAF pass gave this:
Again, in the past, the ASIAIR would have stopped at the third green dot from the right, because the 4th measurement showed a larger star size, and would have been wrong.
They are still using an unconstraned second order curve, though, and the curve is not always drawn with the correct Y offset. But that is not important, the important thing is the abscissa of the minima.
If it maintains this kind of consistency (and it is really ignoring second pass star sizes), the autofous in ASIAIR may finally be usable.
Chen
