Hello,

I was wondering if SGP has a good focus routine for smaller objects such as small galaxies ? I found that focuser is slightly off.

Is there any what to improve focus ?

Kind regards

Roger

Hello,

I was wondering if SGP has a good focus routine for smaller objects such as small galaxies ? I found that focuser is slightly off.

Is there any what to improve focus ?

Kind regards

Roger

Add on question:

Does the Hyperbolic curve fitting routine use a 3 or 4 parameter model? A 3 parameter model can result in fit errors where as a 4 parameter model with be more robust.

Mark W

Im not sure how to answer this question. SGPro does not have a focus routine per-target type. There is just one routine. Iâ€™m also not sure how to interpret â€śslightly offâ€ť. In terms of improving focus, it can depend a lot on the actual target. Is it a data poor area? how many stars are found when in focus? And then also on your AF parameters (number of sample points, steps between sample points, minimum star size) and finally based on gear (focal length, mirrors and focuser backlash).

If you want to attach logs of what â€śslightly offâ€ť focus looks like, we can try to help. Otherwise, there is this:

https://help.sequencegeneratorpro.com/UnderstandingAutoFocus.html

Hello,

Thank you for answering. Could you give me an example of a 3 or 4 parameter model?

Kind regards

Roger

Hello Ken,

Ill upload at the logs.

Kind regards

Roger

The general equation for a hyperbola is:

(x - x0)^2 / a^2 - (y - y0)^2 / b^2 = 1

x0 and y0 are offset parameters. In this case, y is the focuser position and x is the HFR to orient the hyperbola properly.

A 4 parameter fit would estimate a, b, x0 and y0 to find the minimum error between the data and the model prediction.

For a 3 parameter model, the position offset, y0, can be estimated for the 7 or 9 data points in the focus run and a least squares fitting routine can then fit 3 parameters, a, b, and x0.

The 4 parameter model would be the most accurate, but can be more difficult to find a viable solution to fit the curve to the data. It would be slower than fitting a 3 parameter model.

See James Lambâ€™s YouTube channel for a recent video on his testing of curve fit routine for auto focus in NINA.

Mark W

1 Like