Why no one in NINA is looking into developping a single star autofocus instead of current local field method

Auto-Focusing
19 replies•479 views
sebaromano avatar
sebaromano avatar

I am surprised no developpers in NINA are looking into implementing a single star autofocus method versus the current one which average the HFR of several stars in the image field.

You can find the single star autofocus method in Voyager which was the primary reason why I was using Voyager and it is quite brilliant, effective and fast.

I am not a software developper so I cannot say how difficult or easy is to develop.

And perhaps I should ask all above questions into NINA discord channel …and probably I will but I was wondering if anyone knows anything already :)

Sebastiano

Well Written Respectful Engaging
Arun H avatar
Arun H avatar

Probably because Voyager has a patent on it?

Tony Gondola avatar
Tony Gondola avatar

I think it’s a choice of chasing HFR or trusting in statistics which is likely more effective across a wide range of apertures and conditions.

Jeffrey Kieft avatar
Jeffrey Kieft avatar

sebaromano · Mar 24, 2026, 07:00 PM

I am surprised no developpers in NINA are looking into implementing a single star autofocus method versus the current one which average the HFR of several stars in the image field.

You can find the single star autofocus method in Voyager which was the primary reason why I was using Voyager and it is quite brilliant, effective and fast.

I am not a software developper so I cannot say how difficult or easy is to develop.

And perhaps I should ask all above questions into NINA discord channel …and probably I will but I was wondering if anyone knows anything already :)

Sebastiano

I’m not sure what the advantage would be. Using multiple stars gives statistics across the entire field, or using the subset of stars that you want to use, and plug-ins such as Hocus Focus allow even more robust analysis. Using many stars has a lot of advantages and seems more robust and flexible. I am not sure I would trust the analysis from a single star.

Kyle Goodwin avatar
Kyle Goodwin avatar

Arun H · Mar 24, 2026 at 07:26 PM

Probably because Voyager has a patent on it?

It is not patented and there are very similar focus systems which existed long before Voyager’s implementation, such as CCDWare FocusMax. It’s not an intellectual property restriction that results in it not being in NINA.

Helpful Concise
Noah Tingey avatar
Noah Tingey avatar

Is the whole-field autofocus analysis not already essentially instant on any modern mini PC? I’m not sure that I’d want the downsides of single-star autofocus just to save, like… 0.15 seconds of computation time.

Well Written Engaging
Arun H avatar
Arun H avatar

Kyle Goodwin · Mar 27, 2026, 12:27 PM

It is not patented and there are very similar focus systems which existed long before Voyager’s implementation, such as CCDWare FocusMax. It’s not an intellectual property restriction that results in it not being in NINA.

Kyle, you are probably right. I did a search on Google patents and there is nothing in Leo’s name though that does not preclude that he is licensing a patent.

Whatever Voyager is doing seems to be working well for its users, since the speed of Voyager’s autofocus is frequently cited by its users as a differentiator. And, given some very capable imagers are using it, there does not seem to be a deterioration in quality versus whole field autofocus which NINA does.

Helpful Engaging
Reg Pratt avatar
Reg Pratt avatar

I used Voyager for years and have never found either if its AF routines to be any faster or more accurate than NINA. The only time single star AF is beneficial is if your field isn’t flat and theres a huge HFR difference between the on axis and off axis stars. If that’s the case you could probably use the RIO feature to greatly narrow the amount of stars NINA uses for the run.

Helpful Concise Engaging
sebaromano avatar
sebaromano avatar

Reg Pratt · Mar 27, 2026, 10:33 PM

I used Voyager for years and have never found either if its AF routines to be any faster or more accurate than NINA. The only time single star AF is beneficial is if your field isn’t flat and theres a huge HFR difference between the on axis and off axis stars. If that’s the case you could probably use the RIO feature to greatly narrow the amount of stars NINA uses for the run.

That is a very good explanation and point.

Thank you all for the inputs. All valid :)

Respectful Supportive
Willem Jan Drijfhout avatar
Willem Jan Drijfhout avatar

Just to clarify, the single star autofocus in software like Voyager or FocusMax is not just selecting a single star in the frame. It will slew to the nearest star that falls within a certain (customisable) brightness range. There it will run through one half of a V-curve and fit a straight line based on multiple stored V-curves that are created during the calibration process. As it only measures one star, it only captures/downloads a very small area of the frame. So it typically rattles off this curve in no-time, hence the gain in time. Slewing and re-centering of course work the other way around and slow things down.

I often use the traditional whole field multi-star field curve. I use a generous exposure time per sample point, so it is definitely slower, but it gives me a V-curve to look at in the end which I like. But if you have a small galaxy or cluster in the middle, HFR values in the center of the frame are always far off, so probably better to use the single star method. Even though this gives you just a number and a star profile at the end, not a curve. A strong point of the single star method is that each autofocus run is analysed based on a series of calibration runs, so it models the system, rather than starting from scratch each time, where outcome can be influenced by conditions.

Well Written Helpful Insightful Engaging
Arun H avatar
Arun H avatar

Willem, thanks for the explanation. This really helps understand. What I cannot see if why it would be faster, other than it just downloads half the points or less that are needed. From a download standpoint- the cameras I am aware of do not give you the option of selecting a region of interest, so the download speeds should be the same. Processing will be faster but that is not really what causes a slowdown.

Concise
Tony Gondola avatar
Tony Gondola avatar

I suppose pros and cons could be argued back and forth on this and time taken from integration can be an minor issue but which approach gives better results across all imaging systems or is the only difference time?

Alex Nicholas avatar
Alex Nicholas avatar

Same reason nobody does single star guiding instead of multi-star anymore. Taking a field average is FAR superior than a single star..

Arun H avatar
Arun H avatar

Alex Nicholas · Mar 28, 2026 at 02:00 PM

Same reason nobody does single star guiding instead of multi-star anymore. Taking a field average is FAR superior than a single star..

I would not be so quick to dismiss Voyager’s AF method. Honestly, at this point, I’d like for actual Voyager users to chime in with their experiences using it.

Well Written Respectful Engaging
Dark Matters Astrophotography avatar
Dark Matters Astrophotography avatar

The big draw to the single star focusing method is for focusing narrowband filters, IMHO. voyager has field based focusing as well, which works great for LRGB, but for narrowband work I only use the single star method as I find it gives the best results in a fairly quick runtime.

Helpful Concise Engaging
Bill McLaughlin avatar
Bill McLaughlin avatar

Kyle Goodwin · Mar 27, 2026, 12:27 PM

there are very similar focus systems which existed long before Voyager’s implementation, such as CCDWare FocusMax.

Indeed. That system has been around since the early 2000s. The developers (Larry Weber and Steve Brady) were present at a conference in Salem Oregon in 1999 where there was much discussion (and one early-days presentation) about focus. They decided that an automated focus should be doable and within a few years, FocusMax had been developed and it was released at the same conference a few years later.

I was present at those conferences (called “Imaging the Sky”) and was the one who gave the presentation on manual focus at the 1999 conference. That was very primitive stuff by today’s standards.😊

Well Written Helpful Insightful Respectful Engaging
John Hayes avatar
John Hayes avatar

Back in 2017, I presented a workshop on precision focusing at AIC. As a part of that presentation, I made some lab measurements using a couple of C14 Edge telescopes to determine the uncertainty inherent in the “V-curve” algorithm when using a single star—and it turns out to be pretty good; but, only when you consider the effects of seeing. Remember that you want the focus to be set within ± ½ DOF, which is about half of the uncertainty of a single star V-curve. Using N multiple stars decreases the uncertainty by 1/sqrt(N). So, using just four stars reduces the uncertainty by a factor of two to produce the result that you want. The only disadvantage of using multiple stars is if the telescope has a curved or tilted focal surface so that stars over the field come into focus at different positions of the focuser. The second best way to handle that situation is to be able to specify the field to use for focusing; not to select fewer stars. Of course the best way to fix it is to fix the optics! Implementing single star autofocus for NINA would be a step backwards and that’s why no one has implemented it.

📷 image.pngimage.png

📷 image.pngimage.pngJohn

Well Written Helpful Insightful Engaging
sebaromano avatar
sebaromano avatar

Excellent explanation and thanks for contributing in knowledge sharing via this post 😊

Respectful Supportive
Georg N. Nyman avatar
Georg N. Nyman avatar

Very interesting explanations!

I use Voyager and NINA and from my experience - but please do correct me, if this is a false impression, the result depends also very much on the choice of AF hardware brand and the optical system.

I have the ZWO EAF, Pegasus Focus Cube 3, iOptron iAF and the Primalucelab SestoSenso 2 ( “collected” of several years). AF on the same optical system gives more or less (more less) similar but by far not nearly identical results. I tried them all on my RASA11 (f/2.2, 620mm fl)

The least accurate from my experiences is the Pegasus Focus Cube, then comes the iOptron iAF, the EAF from ZWO gives reasonably repeatable results but by far the most accurate and most reliable results come from the SestoSenso 2 from Primalucelab (however that one is a bit a diva - sometimes, it does not connect, sometimes it needs to be reconnected).

I think, those results are caused by the step size, the repeatability of the mechatronics and the optical data of the system.

Having tried the same systems on my 12” TrussRC (f/6.8, 1950mm fl) - all of them are reasonably well usable.

CS
Georg

Well Written Helpful Insightful Respectful Engaging
John Hayes avatar
John Hayes avatar

Georg N. Nyman · Mar 29, 2026 at 06:53 PM

Very interesting explanations!

I use Voyager and NINA and from my experience - but please do correct me, if this is a false impression, the result depends also very much on the choice of AF hardware brand and the optical system.

I have the ZWO EAF, Pegasus Focus Cube 3, iOptron iAF and the Primalucelab SestoSenso 2 ( “collected” of several years). AF on the same optical system gives more or less (more less) similar but by far not nearly identical results. I tried them all on my RASA11 (f/2.2, 620mm fl)

The least accurate from my experiences is the Pegasus Focus Cube, then comes the iOptron iAF, the EAF from ZWO gives reasonably repeatable results but by far the most accurate and most reliable results come from the SestoSenso 2 from Primalucelab (however that one is a bit a diva - sometimes, it does not connect, sometimes it needs to be reconnected).

I think, those results are caused by the step size, the repeatability of the mechatronics and the optical data of the system.

Having tried the same systems on my 12” TrussRC (f/6.8, 1950mm fl) - all of them are reasonably well usable.

CS
Georg

George,

You are right! There are can be a lot of variables that affect how any particular system might perform and in my AIC workshop, we went over a lot of that stuff—ranging from the effects of backlash, hysteresis, and the atmosphere. That’s why the data that I took using the Edge (shown above) used an air path of only about 6”. That eliminated any possibility of seeing effects so the results were only dependent on the mechanics. Under the sky, I see a lot of variation simply due to seeing so that’s a major issue in determining the ultimate precision that any system can achieve. Poor seeing always decreases focusing precision; but, you already know that!

In any case, you can measure the repeatability of your own system. Just pick a night with pretty good seeing and refocus 10-30 times and record the focus position. You can then look at the distribution of the data to determine the standard deviation about the mean. Just looking at the shape of the distribution can also be instructive. It should look like a gaussian curve but if it doesn’t, something might be wrong.

John

Helpful Engaging