Collimation aid in the planetary live stacker

[Borodog]

I have brought this idea up before but I think its time to reiterate and flesh it out a bit.

Currently, I use the planetary live stacker for collimation. I center a focused star (or Jovian moon), engage the live stacker, and evaluate the live stacked and sharpened diffraction rings visually, and adjust the collimation screws to eliminate any asymmetries.

However, if the seeing isn't great, this task can be difficult, even with the live stacker, mostly because it becomes difficult to visually judge asymmetries. I would like to propose a very simple method to make this more objective and less subjective. I would like to have a simple Star Collimation checkbox somewhere in the planetary live stacking tool. When clicked on this would display two dots, one red, and one green on the stacked image. The green dot is simply the COG of the stacked image (subject to a threshold to exclude the background). The red dot is the COG of the square root of the stacked image, subject to the same threshold (before the square root is taken). The square root applies a nonlinear curve (similar to gamma correction) to the image and will hence exaggerate asymmetry and offset the COG relative to the linear image. When the scope is collimated, the red and green dots should be coincident at the center of the Airy disk. If the scope is out of collimation, the red and green dots will be spread apart along the direction that the collimation should be adjusted in, the direction of flaring. A scaling can be applied to the position of these COGs in order to exaggerate collimation errors and make them easier to see. For example a scaling of 10:1 could be applied, wherein an offset of 10 pixels actually represents an offset of only 1 pixel.

This tool would also work with a planetary moon, as long as it was the only object in the field. It would also work with near-focus collimation methods ("inside out" Airy disks) in addition to in-focus collimation. It would work for poor seeing where the diffraction rings are not even visible, simply by eliminating asymmetries/flaring. It will also work for scopes that display trefoil/trilobal first diffraction rings (like mine).

As with any tool, the user would have to take care to use it properly. The user would need to make sure the scope is in equilibrium, as a thermal plume will cause an asymmetry/flare that would affect the location of the dots.

Thanks for your time and consideration.

Cheers,

Mike

[admin]

Hi Mike,

thanks for reminding me of this one - interestingly I already had a sticky note on my desk labelled 'Collimation' today following some discussions I had at an Abingdon Astronomy Society meeting last night, so both of your suggestions are timely.

While the square root would introduce a non-linearity and give some offset as you describe, I wonder if a thresholded pair of measurements would give a clearer signal - ie.

Bright peak center - center of mass of everything about 30% pixel brightness
Rings center - center of mass of everything between 10 and 30% pixel brightness

The percentages are just values made up on the spot, but you can see the idea that this approach would hopefully exclude most of the contribution contribution from the peak towards the calculation of the second point.

Some sort of automatic method to recenter the star once a collimation adjustment is made would be cool too... Might be too hard unless you have plate solving data for orientation/scaling

cheers,

Robin

[Borodog]

Robin,

The main goal in my method is simplicity; the idea being that simple (and useful) ideas have a better chance of being implemented.

I am not sure that the threshold method you describe can work for near-focus collimation, where most of the light is in a bright ring surrounding inner fainter rings and the central spot. It might, but I’m not sure.

I use Feature Tracking to keep the star centered. There’s no way that plate solving would be of use; the collimation star is the only object in the field.

[admin]

Hi Mike,

if you already have plate solving information, that gives the mapping between pixel location in the image and sky co-ordinates, which means that even with just one target in the field, you can recenter it because knowing it's current position lets you work out the pixel offset and therefore the co-ordinate offset to center it. Anyway, as you say, that's perhaps too far at the moment.

Could you grab a couple of short videos the next time you do this? Perhaps one 'before' adjustment with the outer rings offset and one after - I can then see if I can get anything useful out of them. Whenever I am collimating under UK conditions, the random variability due to seeing always seems so large that I wonder if it would even be possible to get a usable signal from image processing...

cheers,

Robin

[Borodog]

Robin,

There’s no plate solution possible at 2800 mm f/10 with any of my cameras.

I will definitely grab some videos for you the next time I collimate.

Mike