Good vs. Bad VCurve?

Multiple VCurves -which ones to say yes and which ones to say no or delete for best Autofocus results in calculating a RoboFire V curve. I have studied the manual and the Wiki. The Wiki says that autofocus can be improved by making more than one VCurve. Looking at my table of VCurves what comprises a good VCurve that should be checked “yes” vs. no” to use in the calculation. I know what the data is showing but not what it means as far as how to interpret which is the better curve(s) for the best autofocus result. For example do I want the lowest or highest PID? Do I want the left and right steps HFD to be as close to mirror images of each other? Or do I choose the curves with the smallest step increments?
My related question is once the actual visual curve is no longer shown on the screen is there a way to have it display again from the table of individual Vcurve data?
Leonardo has been very helpful and responsive answering my new user specific support questions but I thought this was a good general forum question coming from someone new to VCurves and truly automated focusing.
Thanks so much,

Dear Bob,

we have talk about in this forum but happy to answer.

Vcurve if your focuser work fine and without backlash or slippage and your resolution is on human capability its all equal so nothing to do, just 3/4 vcurve are enough. I see this on 90% of user running VCurve.

To decide the vcurve when things are different you cannot only based on the number in the table but you mus see the vcurve graph during the realization and see how V shape is your vcurve and decide after to remove. If have an horizontal series of points (backlash or cable retains or slippage) , if one side have a total different slope pendence, if seeing ruin the running with outliers points: this is a curve to remove.

Looking at table you can only decide to remove the vcurve having a slope left/right value really different from the average. In a real world this will be not possible (ideal vcurve = equal slope value with different sign). Different value of slope right/left from average means a bad vcurve, if the resolution is so high this diffrence can be amplified in this case more vcurve are need to have a good average.

Before delete always use the yes/no mechanism, in this case you can reinclude the vcurve if needed. Press the recalculate button and apply.

PID is something in second order to evaluate, depends on your focuser resolution the difference so better to not evaluate. Also because it is a direct result of slope difference.

What means slope difference : uncollimated instruments, tilt of sensor / focuser tube, particular correction system like Petzval can amplify this difference

In all of this there is the AI engine fixing things for you during the real autofocus.
I hope this helps.

All the best

Thanks Leonardo. Very helpful information verifying what I had assumed. I did notice that the graph being realized for my latest VCurve, after having made a few settings adjustments, showed the points on the curve progressing in a nice straight 45 degree line on both sides with no horizontal outliers. So if it looks pretty and uniform then basically it is good. :+1:
All the best,

1 Like