When I did test on auto focus, I used area around polaris, when I found there was an issue in auto focus.
Basically when it was farther out from focal point, there were not many stars and polaris was a perfect candidate for HFR calculation. When it was near focus, there were many more faint stars so polaris were not included.
Since polaris was very bright, its HFR was way bigger than faint stars, also because # of stars were much less when out of focus, so including polaris skewed the V curve drastically.
It would have been Ok if polaris were always included, but it was not the case. It was included when focuser was racked out, but not racked in, which created a asymmetric V curve, and resulted large error in focal point calculation.
So is there an option to exclude stars over certain HFR?
My experience, but as a beginner: Autofocus can fail if a very bright star is in the FOV that dominates the FOV (Vega for example). And it can give less than optimal if the FOV contains only faint stars that do not register when the initial defocus is performed and no star registers a HFR. I try to pick FOVs that give a clean “V” and avoid FOVs with either bright stars or only faint stars. Then I center on the target. Also, your steps are important to success, I fiddled with mine quite a bit to home in on acceptable values.
@UlteriorModem As @ewiley explained in detail, this seems a real issue with “average HFR” approach. In this case, a single star HFR would be preferred.
@ewiley The problem here is that it can not be automated, which is not feasible because it is more like a manual focus.