For several months I have been noticing a drift issue in my images. When fast-blinking it seems that my target has a tendency to move more to the right before meridian flip, and more to the left after.
Here are 2 short clips of the two latest sessions to illustrate my point:
The first is a blink through 214 subs @60s in L (spanning 4.5 hours). This was already after meridian flip, so one can only see the tendency of object drift to the left.
https://youtu.be/oyzA29KQa5w
The second is a blink through 85 subs @300s in OIII (spanning 8.5 hours). Meridian flip is happening sometime inbetween and one can clearly see how the drift direction changes right after. The drift is not constant, sometimes it is 'coming back', but the tendency is still obvious.
https://youtu.be/jLdAm8puvng
For L I dither every 5 subs, for OIII I dither every sub - both at 10pixels. ASIAir + guidescope setup.
In the L example one can see the 'non-dithered' frames being rock solid, and movement only introduced every 5 subs - when a dithering step is performed. Also, I polar align separately for every session as my setup is non-permanent. Thus, I rule out polar alignment as root cause. The remaining ideas are flexure or an issue with the ASIAir dithering algorithm.
This problem has been observed over many months, several different ASIAir SW versions and the bias towards a specific direction is reproducible. I have also started a larger thread about it on the Cloudy Nights forum which I can link here if required.
Bottom line: Lots of voices pointing towards the implemented dithering algorithm in ASIAir being the culprit. There are easy ways to optimize this apparently if ZWO is willing to revise and potentially update the code. I will happily share the code proposal here which a CN user has kindly provided.
As an alternative solution, could ZWO consider to add an option 'recenter after x frames' which is a standard option in e.g. NINA?
