Focus or Guiding causing high FWHM?

I am revisiting some of my astrophotos to clean them up for publishing. In many cases it’s becoming obvious that my stars are not as sharp as they should be. I struggle trying to determine whether the issue is focus or guiding.

I always focus using a live 8-bit video stream and autofocus to yield the smallest stars. When I look at my photos later I notice that the FWHM values vary from night to night and yet, in comparison, my guiding is fairly consistent. If FWHM varied with guiding then I would blame guiding. If guiding is consistent and FWHM varies I blame focus.

I autofocus using my own software It is designed to use a live video stream from any other application. In my case it is using the Liveview from my Nikon D5300 being displayed by Digicamcontrol. All I have to do is position my mouse over the star in the live stream and then press the spacebar to lock the region of interest (ROI). The square of pixels in the upper left corner is the ROI. The green bargraph beside it is the star brightness which alerts me if the star is saturated.

I calculate a value called Star Size which is loosely defined simularly to FWHM. The Half Maximum value is determine for the star and then all of the pixels that are brighter than this Half Maximum are counted up and displayed as the Star Size. The half maximum value is recalculated for every live video frame and the star size values are subjected to some averaging which gives the star size value some sub-pixel resolution.

As the program sweeps the focuser position thru a range of values the star size values tend to form a parabola. When the sweep finishes the program curve fits the data to a parabola, displays the parabola and re-position the focuser to the minimum value of the parabola.

Two observations from the video are:

1)The Optimum Position starts off at 18211. Then it changes to 18198 and finally to 18182. This shows the variability of this approach. It’s important to note that this is using the Moonlite high resolution stepper motor focuser on my BK80ED refractor. I had hoped that by curve fitting the data that the variability would be minimized.

2)After each AutoFocus you can see the red dot bounces around a fair bit. This tells me that the Star Size calculation has a lot of variability. It could also mean that the seeing is really causing the star to dance. I was hoping that the averaging would quiet this down.

I have been reading up on the Half Flux Diameter (HFD) calculation. This calculation is more involved than FWHM but it may also be more stable.

My Starsize calculation is calculating the area, in pixels, that contains pixels whose values are greater than the Half Maximum. This is basically

Starsize = PI x R^2 where R is the star radius and 2 x R = FWHM

HFD first requires that a centroid calculation be performed to determine the star’s position with sub-pixel accuracy. Next, two sums are calculated. First, all of the pixel values are summed together. Secondly, a radius weighted sum of all of the pixel values is calculated. This just means that the pixel values are multiplied by the distance this pixel is from the star’s position before they are summed together. Finally, the radius weighted sum is divided by the pixel sum to yield the Half Flux radius. This value is multiplied by two to yield the Half Flux Diameter.

The HFD is a more complicated calculation but I think it will be more accurate for a few reasons:

1)HFD uses ALL of the pixels instead of just the bright pixels.

2)I know that the centroid calculation uses a position weighted sum and PHD2 uses this calculation to achieve very high accuracy.

So…on my TODO list for this winter is to modify this program to use HFD instead of Starsize.

Also this year I think I was more focused on using the QHY294C than worrying about autofocus so it’s possible that some focus stars were saturated which is a no no.

Always striving to learn something…especially from my mistakes.

Peter