What I have done for autofocus on targets with few stars in the field is to find an area near the target with a good number of stars and create a sequence using the Faming and Mosaic wizard. I then slew to and center on this target and run my initial autofocus. You can then set it as the focus position. I had great success with my SCT to autofocus for each filter (wide and narrowband) on an open star cluster. After setting the focus, I then slew and center on the target, set the camera rotation, and then calibrate PHD2 on a guide star. With target and focus positions set, then autofocus will work flawlessly for me. After initial focus on the star cluster or area with enough stars, good focus is achieved and refocusing on the target is usually successful, provided the transparency and seeing are sufficient, even with few stars and with Ha or Sii filters. No plate solving needed other than for recentering.
To automate the sequence, you can move the focusing sequence to the top of the list, set the camera rotation to that of the target, do the autofocus and let the camera reach set point, take one exposure to save the HFR on the imaging binning setting, then move on to the first imaging target.
Has autofocus been enhanced so it no longer picks up double stars, small galaxies and other non-star objects that throw off the HFR calculation as discussed in this thread?
I didnāt see this mentioned in recent release notes.
Not that I am aware. By measuring multiple stars, however, single anomalies have little impact on the result.
The one thing it really doesnāt like is globulars. It confuses the calculation.
Markās prior comment echos one of mine from a while back. There is a focus target feature in SGP but it seems to have been forgotten about over the years. Markās idea is what I expected the focus target to implement. I should make it an official feature request.
ISTM that the ability for the user to specify the focus stars to use would answer most of the questions and issues raised here; I could click on 10 or so things that I know to be single stars, and only they would be used for autofocus.
Or alternatively, allow the user to specify an area (probably a polygon) inside of which it should NOT use any objects. That would let you (for example) draw a box around M42 and make it only use objects outside the box.
I see similar problems in the focus routine (RC8@1625 mm FL, ASI1600MM) not producing stable results in the presence of small, extended objects. Specifically, the focus curve becomes asymmetrical. This is not a problem in āstar onlyā fields. Of particular impact are relatively bright galaxy cores. I, too, think a platesolve-based approach would yield the best results. But even outlier-rejection might improve this to a better outcome.
I suspect that using the same stars throughout the autofocus series would be a bit more resistant to variations in seeing, which would help me a lotā¦
Very many people use SGP because it can run unattended. Having to choose a set of stars at each AF run does not seem like the way to go. Hopefully with the new ASTAP approach in v4 (because it calculates a far more statistically robust HFD/HFR), you will find a dramatic improvement. In testing it has so far worked very well for me.
SGP does not currently have an interface to PW for AF so this question is currently academic.
If you read earlier posts in this thread you will see that some SGP users suspect that in SGP V3 the AF routines are including within the set of star detections used for its HFR calculation, sufficient numbers of non-stellar objects (galaxies, planetary nebs, cosmic ray trails etc) to cause a significant displacement of the calculated best focus position from the position that would be calculated if such objects were excluded. The idea of using platesolve on the AF images is that SGP (or PW) would compare the candidate list of AF star detections against the underlying platesolve star catalogues to eliminate detections of non-stellar objects from the HFR calculation thus leading to a āpurerā result.
FWIW, at the moment I am undecided on the need for this added sophistication:
a) I havenāt seen any actual analysis to show that having non-stellar objects included within the HFR calculation is by itself leading to a significant displacement of the calculated best focus from the ideal.
b) There are in my opinion other causes for displacement of calculated AF position from ideal that are more likely leading to poor results. For example, it is acknowledged that using parabolic curve fit can give a poor result if the AF curve is not reasonably symetrical about its axis and in particular if one wing of the curve extends well beyond the other. This problem can be ameliorated by raising the required quality level of the curve fit so that an AF rerun is triggered unless fit quality is good. Also there has been a suggestion that v3 SGP HFR method was maybe not correctly integrating the flux level for individual star detections, which would lead to false HFR values in AF calculations.
c) In SGP v4 beta release, SGP has added a new AF metric provider, ASTAP. I have been testing this carefully and think it gives much better AF results compared with old SGP HFR method. Having said this, I am using a modest refractor scope so my good results may not extend to say users of long focal length refractors. For info, ASTAP applies some tests (eg eccentricity) to eliminate non-stellar objects from its star detections. Also the HFD result is median value so abnormally large candidate detections such as galaxy cores should not unduely influence the calculated best focus position.
d) I have done a little research but so far have not managed to find anything on PW web-site that confirms that PW AF is actually using platesolve to eliminate non-stellar objects from its HFR/HFD calculation. Maybe it is only using algorithmic techniques such as ASTAP is using.
I donāt know what software you are using but I found recently that ACP Expert (available on 60 day trial) does have interface to PW for Autofocus. Long-term purchase is however an expensive option.
Thanks for the detailed reply, Mike. Iām struggling a bit with focusing using an Optec SWX30 focuser, for which I get basically no (or two-week-delayed) āsupport.ā I recently got HFR readings well under 2 and I thought I was good to go, but then SGP (or the focuser or whomever can āloseā such settings) lost that focus, and during that same session last night I couldnāt get anything under 5 and I was not able to achieve acceptable focus to run the sequence with using the focusing routine. Wasnāt an issue of clouds or wind or anything that I could determine. I need some better strategies for obtaining good focus as things arenāt quite as āautomaticā as I had expected them to be when I set up the rig this fall.
Iāve read many dozen comments but havenāt yet seen any tips or tricks or strategies suggested, so if you have any of those or know of any links to good techniques, Iād love to hear about them. Iām using a Skywatcher Esprit 100 ED on a Skywatcher AZ-EQ6 Pro mount for what its worth.
If nothing else has been done to change the focuser position then it could just be that there has been a temperature change sufficient to markedly change the focus position since your last time out.
The AF method often does not work well if the scope is not close to a good focus at the start.
When I was first starting with AF I would use the Coarse In/Out buttons on the Focus Control tab, see RHS in below screen-shot, to move the focuser. Take an image after each move (Take One on Frame and Focus Tab) and you should be able quite quickly if focusing is improving or not - change direction as appropriate. Use the Fine control when you get close to your HFR 2. NB. My step size values may not be appropriate for your rig. Adjust as needed until you start to see a decent change in HFR.
If you have not already done so I would also check what distance your focuser needs to move to get from best focus HFR to approx 3 x this value. If you are using 7 steps for focus your step size should be about 1/3 this distance. If AF only detects 1 or 2 stars then I suggest you reduce the stepsize so you have at least 4-5 star detections. You might also increase your exposure time to see if that gets better data for the AF routine. Once you have things going nicely you can decrease exposure to speed things up.
You donāt say if you are using filters but you will need longer exposures for RGB filters compared to Lum. I also found I needed to decrease the step-size to get focusing for RGB filters to work well as there will be fewer star detections with these filters compared to L.
Hope you soon managed to get things working smoothly.
Thanks, MIke. Quick question - Does SGP include a pointing model, like T-Point?
Also, I"m using an OSC, so filters are not currently an issue (planning to use LP or narrowband when everything else settles into place.)
And good tip about Take one - I use it on the continuous setting when I am manually adjusting focus. My step size is usually 50, and the exposure depends on what I am pointing at.
So many variables!
Thanks again.
SGP interfaces to mount software that use a pointing model but does not have one itself.
If you have continued problems with AF I suggest you tick the option to save AF packs and include these with your problem report.
And I assume you have checked for possible backlash. The characteristic sign of a problem is a limp curve on the RHS of the AF curve. Plenty of other posts on this topic and how to address it.
