Taks and Autofocus, again

Ken and Jared,

please hear us again, the suffering Tak users, who only ask to introduce selectable rectangle zone to exclude center stars from calculating HFR during autofocus. This has been asked for years now and it was on your back log since 2017 :frowning:
Another perfect night gone to drains, waste of time, gas, energy and it really getting us frustrated.
Perfect guiding , perfect Vcurve , constantly providing elongated stars at corners, which always can be fixed with manual adjustment of focuser steps.
This time i had my friend beside me, me running FSQ106 with RD645 reducer and him FSQ85 with QE73. Both of us runs small ICX694 sensors, which are definitely should be covered by these reducers.
Yet, you can see how perfect Vcurve translated into FWHM and Eccentricity around the field.
We need this fix! Please hear us, crying out loud Tak and SgPro users!

Donā€™t let us look for alternatives and abandon SGPro, we are loyal usres!

eccentricity
FWHM

1 Like

I think you are suggesting the same thing that NINA does, which measures star FWHM in a rectangular annulus. I donā€™t see this problem on my Tak85 but have seen others with strange middle bumps near focus, making Tak V curves more like a W curve. Probably worth reposting this idea as a feature request - even for non-Tak owners, when there is a slightly imperfect flat field, finding a median focus point is better than a perfect one for central starsā€¦ as the brain sees the high resolution in the middle and judges the others in comparison.

Gotta say that this must one of the following:

  1. Paying attention to numbers as opposed to appearance. Does the final image really look that bad?

  2. The reducer. Do you see this w/o the reducer?

  3. Just a bad individual scope.

I have been using SGP with an older fluorite FSQ 106N for years and have not seen this issue at all, at least visually, when using an 8300 camera - and I am legendary with my imaging friends for being fussy about round stars. The 8300 is, I believe, larger that what the OP is using. OTOH, I have never used a reducer with any of my Taks (I also have a TOA-130).

Finally I have recently used the 106 with a much larger 16200 camera and in the process was able to try it on both my 106N and the latest 106EDX. Surprisingly, the older 106N was significantly superior to the newer version in terms of flatness (on the much larger 16200 chip). Whether this is generically true or was just a bad example of the newer EDX it is hard to say but I have heard numerous reports of highly variable quality.

Of course with the larger chip I was able to see, just barely, some corner elongation, but that was not unexpected.

Probably not what you wanted to hear but I have been imaging since the mid 90s and have some small expertise.

Having said all of the above, I still support the idea of a user defined focus area since this can be useful for other purposes as well (such as avoiding objects that might confuse the system).

@AstroScience

Hi. May I just ask what version of SGP you are using, and if v4 Beta, if you are using ASTAP focus metrics?

My understanding is that when using the ASTAP focus metrics option the HFD values used by SGP during AF are in fact the median value of the selected AF stars (but note unless there has been a recent change I do not think that these stars are absolutely the ones highlighted in the image display during AF - which display uses the old SGP detection method.

Therefore, unless you have heavily cropped the AF window (under autofocus options) so that AF is using only the most central stars, I would personally expect the focus to be determined by the lowest median HFD measurement of the stars across the whole image rather than median of only the most central stars.

Out of curiosity can you give an indication of the delta (microns if possible) between AF using only the central stars and whatever position you think gives the best overall focus?

If you have saved any of your AF images what HFD values do you see reported by ASTAP CCD inspector?

Regards

Iā€™m still on 3.2 version.
Guys, just use FSQ and Autofocus words for search and you will find at least dozen threads explaining the issue and requests for autofocus ROI or offsets. Autofocus produces perfect curve, but after it you need to tweak the steps of focuser manually to get the round stars in corners.
Takahashi suggests to focus on star 2/3 out of center, they probably know what they asking right?
The large pixel cameras will swallow all the issues most of the time, therefor you may not have these problems and no complains.
And again, it wasnā€™t request to troubleshoot my issues, itā€™s just another request to let us this option to exclude central area and use stars close to borders for focusing, exactly as NINA implemented it.

I do not disagree with the request but there is a distinction between single and multiple star HFD optimization though. Takā€™s advice is likely to be aimed at those who do single-star focusing.
I have the BabyQ and the AF produces very good results. SGP uses an aggregate HFD measurement of the entire frame. If you consider a 3x3 matrix, the central third only represents 1/9th of the entire star count and so has little influence. Excluding a middle core might not be the silver bullet you were wishing for.

Buzz, Takā€™s advise is aimed for all FSQ users, meaning, if you want to get optimal result with round stars across the chip, you need to use single star for focusing placed at 2/3 from the center. Same advise applies for focusing fast Newts for example. While using multistar/full field focusing, this will average all stars, not only those located at 2/3ds from the center, but average from all throughout the field, which is obviously is not optimal
Also, you may have 6-9 um chip , or use binning, which will ā€œswallowā€ all those problems at corners with nonperfect focus and you may not see the problem at all, therefore ā€œI have good resultsā€.
And no, i would not think of the field as 3x3 matrix with central third being 1/9th of the entire star count.
2/3ds means that the star should be on the border of ~60% internal area and at ~30% of outside area.
Excluding stars 50-60% from central area for focusing will use and average all teh stars that are at 30% of outside area, bring it closer to Takā€™s advise.
I donā€™t really understand refusal of understanding the concept, as it may also help with other situations, like large glob or galaxy right in center of the frame.
A year ago, it was advised to NINA devs to implement it and it was done in 3 days!
Why we need to wait for more than year for such a simple option, strikes me. When we already have the crop option, can it be simply reversed?
I simply giving up and thatā€™s it.

A couple of things; have you posted this request as a feature request? Previous posts from the developers have said this is the way to ask. It may have been overlooked.

On a personal note, my pixel sizes range from 2.4- 4.6u, never binned and Iā€™m very pickyā€¦ but I also bought the 1.01x flattener for the Baby Q, so it may have a flatter focus plane.

As I said, I think it is a good idea, but not everyone has a Takā€¦ but there is an even more significant benefit of obscuring the middle zone, which benefits most imagers; it fixes the problem of focusing on globs or bright galaxies when they are (usually) centered in the image.

1 Like

Buzz,
you made me smile :slight_smile: Because just 4 months ago you put exactly same request as me in Feature Request category:

Here is another one, from a year ago:

This one old as 2017, describes same problem:

Another from 2017:

Old as 2016:

As you can see, the list can go on and on, so it does got overlooked, but not because it never mentioned anywhere of the forums.

Oh dear. I had a curious sense of deja vu when I did that post !!