Multistar FWHM and full-Moon sky brightness

[Anthony Quintile]

I am new to auto-focusing, so please feel free to offer any otherwise obvious advice.

I was using the Multistar FWHM measurement and frequently getting 20.00 as the FWHM measurement, and today found this quote from another post:

"The value of 20 that you see for the multi-star option is generated when not enough stars are detected to give a meaningful average. If you are seeing that frequently then you should adjust your star detection settings to try to increase the number of stars being picked up by the measurement routine."

I am guessing that attempting to get a good number of measurable stars with a brightly lit sky from the full-Moon, (when I like to learn how to run new equipment since I shoot unfiltered OSC), is going to be much harder than when I have a dark sky?

Generally I was getting inconsistent FWHM measurements with a lot of frames not being used, (I was also experimenting with the single star FWHM function). Is it safe to attribute this to the bright sky?

I was adjusting the black level and other tools, but somewhat haphazardly in an effort to gain some understanding of the functions, but I wasn't getting much difference in results.

Any advice appreciated. I can elaborate on equipment if necessary...

[admin]

Hi,

here are my suggestions for how to set up the star detection and other parameters :

Black Level - look at your histogram – you will see a big peak near the left-hand side coming from the background sky brightness. Set the black level to a value just a bit higher than the right-hand side of this peak so that all of the pixel values that are not brighter than that sky background level are ignored

Noise Reduction - initially set to zero

Averaging - set to 1 frame to begin with

Star Detection :

Min Width - set to 1

Digital Gain - set to off

Min Brightness - 25% (or 10% more than the black level, whichever is higher)

Max Brightness - 95%

Those settings will work for most cases. The reasons that you might want to change them as follows:

Not enough stars being found (but you can see faint stars) - try lowering the minimum brightness or applying a digital gain
Hot pixels are being detected as stars - increase the noise reduction or the minimum star width

Cheers, Robin

[Anthony Quintile]

Thanks Robin-

Having these starting points and understanding should get me on my way.

[chongo228]

Look in the bottom left hand corner and click the star detection tab. It will display how many stars SC can see and is the fastest way to see how your changes affected the focuser. I like to get around 75 to 200 stars (depending on target, light pollution, moon, etc). I'll usually bring my min star size up to 2 or 3 as my last step to bring the star count into range. This will also cut down on noise being counted as stars. I run about 5 second exposures with medium to low gain and collect three samples at each stop during my focus routine.

Once you're done use a mask to focus and compare how SC does with it's focus. Dial in the settings until SC and the mask both say it's in focus. I've got my goto scope dialed in now and haven't had to touch the settings in weeks regardless of target or moon.

[Anthony Quintile]

Following the advice both of you posted (Robin and Chongo228) I am now getting reasonable data every frame. Thanks for getting me pointed in the right direction!

Chongo28, you had mentioned checking results with a mask, which has, in part, led me to a realization, (or misunderstanding). Bear in mind, I am new to autofocusing entirely.

It seems to me that varying and/or average to poor seeing conditions would statistically overwhelm the ability of the sampling to determine truly best focus, at least without averaging a large number of samples and spending a lot of time. This leads me to believe that a well used Bahtinov mask is really a much faster way to determine correct focus. A spike slightly blurred frame-over-frame because of seeing changes can still be seen in the center of the star, whereas a FWHM number that varies dramatically, (more than the difference between focuser position changes), is hard to "smooth out" with a smallish number of samples in a timely fashion and then compare to samples from an in or out on the focuser position.

Am I correct in my assumptions, or am I not using the tools correctly yet? (I could see some potential other issues that I need to average several frames, set backlash, etc.?)

[oopfan]

Anthony,

I know that some people swear by Bahtinov Masks, but I stopped using mine for two reasons: you need a bright star, and you risk accidental loss of PA putting it on and taking it off if you bump the tripod/mount. Also, I focus often, after slewing to a new object and after a filter change.

I use single-star FWHM. I choose a relatively bright, unsaturated star near the center of the field if possible, and set the exposure for approximately one second updates. I begin by turning the focus knob so that the star is out of focus. I let a frame or two pass, and then turn the knob at a brisk pace in order to visualize the V-shaped plot, then stop and reverse direction at a slower pace until I pass through the minimum again. I'll pass through the minimum a couple more times until I am satisfied. The plot auto-scales. That is why I begin out-of-focus and create the V-shape quickly. This dampens the "noisy" plot bars.

Brian

[chongo228]

In my experience a mask might be faster to confirm focus if you have a bright star in the FOV but slower to focus in most situations. Now that I'm confident in auto-focus I don't use a mask to confirm anymore. When I was learning how to dial in AF I would pick a target with a bright star, run AF, and check with the mask (using the bright star in the FOV). My mask and AF numbers were usually within 5 or 10 steps of each other. Remember if the mask and AF focus positions don't match numerically it's probably due to backlash being set incorrectly. Don't worry too much if the step numbers don't line up exactly.

When I started trying to learn autofocus I timed my focus routine with a mask. I'd usually have to slew to a bright star, focus with the mask, slew back to target, plate solve to confirm, and start taking photos again. It took me about 4.5 minutes. Just over five if I was doing NB images.

Now with auto-focus multi-star I can focus in under two minutes without leaving the target. I do three frames of five second exposures at every step. I'll usually do seven steps. I get close to focus by eye and rack out 40 to 80 steps (depending on which scope I'm using) and then send the focuser in for the routine). It's one of the best tools I have learned...probably second only to plate solving.


As far as the FWHM smoothing out you speak of I understand what you're saying but remember it's using the V to calculate the best position and not just each exposure. So each step might only be 15 seconds in the example I gave above but I think it uses all the steps in the "V" to calculate the best position...so any small changes due to seeing per exposure are insignificant when you have multiple data points making the V. You will always have some variance due to "seeing". I could be wrong on this....but that's how I understand it.

[Anthony Quintile]

Chongo228 your help has been outstanding!

I set backlash comp, (44 on my EAF for reference...), and managed to get a curve, repeated with the same results and checked as accurate with my Bahtinov mask. I was doing 5 secs, three exposures, 8 positions, and then returning to best focuser position.

If I am following you, you don't use the function to return to best focus measurement? I haven't looked into it, but I haven't noticed a place to "save" the focus curve to reuse to refocus, and additionally, FWHM would change throughout the night with changes in object altitude and seeing variation, so regeneration of the curve seems to not only be the fastest but best solution. (Returning to best focus measurement also seems to be fixed at 20 steps, which slows that process substantially anyway.)

Is that accurate?

On my initial thinking about seeing affecting FWHM...

It is certainly the case that seeing affects FWHM, but adjusting the star detection parameters and averaging tools seems to minimize this affect. However, it does impact the ability to measure focus accuracy manually in real time. In a perfect world it would be nice to be able to adjust focus 5 or 10 steps and be able to gauge within a frame or two whether that was an improvement or deterioration in focus. Maybe in very stable seeing conditions this would be possible?

Anyway, thank you immensely for your help! I really needed a bit of direction and your advice was perfect.

[timh]

Hi Anthony,

I experience exactly what you have described with respect to focusing under UK skies. Thames valley, winter and under the Heathrow flight path may possibly be about as bad as it gets? For the same reason I don't spend too much time att he beginning trying to optimise focus because it's more like a U than a V and just try to get somewhere near the middle. I also use single star FWHM and also get there roughly and quite quickly just by inspecting short exposure images of stars up at 125% zoom to get them covering about 4 pixels. When it comes to actually taking images I use the SC livestack tool and keep a close eye on the FWHM values of individual frames - Over the course of imaging I continue tweaking focus slightly looking for any consistent fall in FWHM. As the seeing conditions change over an evening I've seen FWHM (pix) values decrease from a barely acceptable 4.5 right down to 2.5 or vice versa -- anyway I get the impression that seeing is really the dominant factor most of the time.

best wishes
Tim

[chongo228]

Chongo228 your help has been outstanding!

I set backlash comp, (44 on my EAF for reference...), and managed to get a curve, repeated with the same results and checked as accurate with my Bahtinov mask. I was doing 5 secs, three exposures, 8 positions, and then returning to best focuser position.

If I am following you, you don't use the function to return to best focus measurement? I haven't looked into it, but I haven't noticed a place to "save" the focus curve to reuse to refocus, and additionally, FWHM would change throughout the night with changes in object altitude and seeing variation, so regeneration of the curve seems to not only be the fastest but best solution. (Returning to best focus measurement also seems to be fixed at 20 steps, which slows that process substantially anyway.)

Is that accurate?

On my initial thinking about seeing affecting FWHM...

It is certainly the case that seeing affects FWHM, but adjusting the star detection parameters and averaging tools seems to minimize this affect. However, it does impact the ability to measure focus accuracy manually in real time. In a perfect world it would be nice to be able to adjust focus 5 or 10 steps and be able to gauge within a frame or two whether that was an improvement or deterioration in focus. Maybe in very stable seeing conditions this would be possible?

Anyway, thank you immensely for your help! I really needed a bit of direction and your advice was perfect.

I use the return to best position option. All of my focusers are on simple refractors with a Moonlite of Feather Touch and I have the backlash dialed in pretty well. I've experimented and the extra time to go to best score didn't make a difference.

You're correct, no way to save focus....you must run the tool from scratch every time. If conditions are stable you could probably get away with shortening the exposures or amount of steps to save time.