Back to the earlier idea of temperature and elevation based focus compensation for an update. Over the last couple of months, I've used the autofocus log to map out temperature and elevation focus position dependencies for my gear. I'll note we should have added elevation to the autofocus log entries; we're going to need that (automation). The path forward to a focus position model seems fairly straight forward.
The idea is pretty simple (shown in Excel for POC). Use autofocus log to plot measured focuser positions vs temperatures with a trendline (function). The coefficients of that function define the focus temperature model. Going further, if you subtract the trend function output from each focus measurement, you create a residual. Plot the residuals set against elevations with a 2nd trendline (function), and you have the additional coefficients for the focus elevation model. Using temperature alone, or temp+el functions, a good/better starting autofocus focuser position can be determined. Longer term, more dynamic options present themselves.
Why bother? Well, as others have noted, if you don't start autofocus from a good starting position (seed), solution time can add up quickly. Worse, a bad seed could escape the solution envelope entirely. To implement this within Ekos, we don't need graphs. We could survive with off-line plot analysis + a couple of radio control buttons (enable/disable temp & el compensation separately). We'd also need fields for the function coefficients (linear + offset for each function could work, but 2nd order coeffs might be desirable for some focusers). The seed should be auto-generated/displayed. It could be enabled for auto use, or just calculated (for manual eval/entry). I attach a couple plots to show the idea. The trendline functions are shown at top/left in each plot (combined for Temp+el at top). The green line at the bottom is a simple calculator for the seed (using temp, el inputs). Long term, and for stable/well defined focuser setups, a dynamic delta el focuser update might allow for longer imaging runs between autofocus events. Comments?