×

INDI Library v2.0.6 is Released (02 Feb 2024)

Bi-monthly release with minor bug fixes and improvements

For those with focus issues

  • Posts: 152
  • Thank you received: 20

Replied by hades on topic For those with focus issues

Hi Ron, what settings are you using for Initial Step size and Out Step Multiple?
I understand your point, that during the autofocus procedure the set of stars in "full field" is changing. Sometimes happens that small portion of a star ,that is too much out of focus, is identified as a small star, with small HFR. This then can lead to incorrect calculation of average HFR of stars in "full field". At least this happens to me sometimes.
Now I have discovered that I have better results if I use Step size = 200 instead of 300, which I used to have before. It is also good to set the Out Step Multiple setting to be half of number of typical iteration in first focusing run, eg: if the first automatic focus procedure takes 13 iterations (from right to left of the chart), I am setting the Out Step Multiple to 6. So the best focus should be somewhere in the middle of the V-curve.
All of this applies to Iterative algortihm
2 years 2 months ago #79799

Please Log in or Create an account to join the conversation.

  • Posts: 348
  • Thank you received: 69

Replied by Giles on topic For those with focus issues

OK, so there are focus issues, we can put these down to:
* We have backlash
* We have slippage
When we say backlash we generally means that there is a delay when we change direction of the focuser that there are a certain number of steps that we take that have no effect, and it is until these steps have expired that focuser actually moves.
When we say slippage we mean that during he course of the movement of the focuser, some movement is recorded where no movement happens at all.

We can measure this, move your focuser from its normal in-focus position 10,000 steps in one direction, and move 10,000 steps back, measure the difference of your starting position to your ending position, this is (more-or-less) a combination of the two added together.

Measuring backlash would be the same process, but with smaller steps (to eliminate the effect of slippage).

Now look at the settings under "Mechanics"

Initial Step Size - this needs to be a number that, not-withstanding the above values, creates a noticeable difference to the recorded HFR in an image.
Max Travel - this needs to be a number that the difference of gets you to within the focus value that you're looking for
Max Step size - this needs to be above the number of "Initial Step Size", if unsure, make it equal to.

20 seconds is pretty high for a focusing frame, the focusing algorithms will be able to detect differences far lower than the eye can detect, you don't need high gain settings, or high exposure settings for focusing, try the similar settings that you use for guiding, only raise them if needed.

You can't focus if seeing is bad, test these under optimal conditions, if they fail due to bad seeing later on, then maybe it was just a bad night for AP.


 
Last edit: 2 years 2 months ago by Giles.
2 years 2 months ago #79807

Please Log in or Create an account to join the conversation.

  • Posts: 225
  • Thank you received: 16
 Peter,

Thanks for the thoughts.  I don't quite understand your first point, as I can see that EKOS is selecting different stars on each iteration.  I believe that this is based on the star size parameters selected in the tabs.  So if a star is selected initially because it fits in the size range... then the size changes because of a change in the focus... it may fall out of the selected stars... which then affects the HFR average without regard to the actual change in focus accuracy.

I do follow the practice of getting the focus as perfect as possible before I begin... and keep the step size and multiple reasonable.  For my 8" EdgeHD I use linear setting of 20 step size and 5 multiples.  It's weird because sometimes I get a perfectly formed curve and other times not.

I do also understand that seeing affects all of this.  Last night the seeing/transparency were both very good... but the auto focus was awful.  I'm going to try again tonight without "full field" option... we'll see what happens!

Again... Thanks!

Ron
2 years 2 months ago #79808

Please Log in or Create an account to join the conversation.

  • Posts: 225
  • Thank you received: 16
Giles,

Thanks!  I've done those measurements and think I've got the right numbers.  I started using linear because backlash is inconsequential... even if it takes longer.  I use 2 seconds for auto focus.  I'm going to try some stuff tonight... will report back.

Ron
2 years 2 months ago #79809

Please Log in or Create an account to join the conversation.

  • Posts: 211
  • Thank you received: 31
Hi Ron,

What do you mean by "only get good focus about half the time"? What happens in the "bad half"?
Sometimes (nowhere near half) I get an especially low HFR value, or even a seeing change so the whole 2nd pass has higher values, and the algorithm often does not deal with that very gracefully.

I also pre-focus at the start of the night. I use the manual single-frame capture button and check the reported HFR. I have a pretty good idea what is right for in focus. If it is higher, then I manually step the focus motor and check again, repeating as necessary. One tip I just learned from this thread is for times when I have changed the setup more drastically I might uncheck full field for the early part of getting to nearly focused, just to speed up the calculation time.

I, too, had wondered about the fact that the stars being evaluated change as focus changes, but Peter's explanation makes sense.
2 years 2 months ago #79810

Please Log in or Create an account to join the conversation.

  • Posts: 211
  • Thank you received: 31
My own question:

Is there an explanation written somewhere for the new behavior of the Linear algorithm, with the "Pending" part? A couple of times I have had that keep moving focus inward (by 1/4 the initial step size) until the curve is decidedly going back up.
2 years 2 months ago #79811

Please Log in or Create an account to join the conversation.

  • Posts: 225
  • Thank you received: 16
Hey Ron (great name btw!),

Last night was a good example.  I had good seeing/transparency and "pre-focused" per your method before starting auto focus (linear).  I got a semi decent curve, but it was far from clear where the bottom of the curve was supposed to be.  EKOS keep trying, but I could tell that it was selecting different stars each iterations and that the resulting average HFR was bouncing around a bit.  Once it completed, I took a 3-minute sub and the stars were clearly bloated (see attached zoom in of part of the sub stretched).

I will say that "half of the time" is a bit of an exaggeration... probably more like 20-30% of the time.  I let my system run over-night most of the time and I have had to throw away several hours of subs after the auto-focus runs later in the night.

I'm also thinking about what Giles said about focusing near a nebula throwing the process off.  Last night I was imaging M1... So maybe a possibility?  I will test that as well tonight.

Like I said... I'm not an expert... but am trying to learn more.

Thanks,

Ron

Last edit: 2 years 2 months ago by Ron Clanton.
2 years 2 months ago #79812
Attachments:

Please Log in or Create an account to join the conversation.

  • Posts: 211
  • Thank you received: 31
Well that certainly fits any definition of "out of focus"! (is that the SCT? Looks maybe out of collimation, as well - but I am *really* not an expert in that realm)
Surely you could see in the focus tab display that you would up with donuts instead of stars (but that doesn't help if you're trying to automate).

When I first started autofocusing I made the mistake of tightening the 'regular' tension adjustment on the focuser, and all kinds of horrible things would happen, mostly related to the stepper motor reporting it had moved but the focuser was still really in the same place (different system, though - I use a Rigelsys NStep).

"Far from clear what the bottom of the curve is" makes me first think to suggest increasing the initial step size. You might (should, really, as long as you are experimenting) also try increasing the exposure time and averaging over 2 or 3 frames - though I still get "bouncy" curves with 2 or 3 frames when the seeing is poor.
2 years 2 months ago #79813

Please Log in or Create an account to join the conversation.

  • Posts: 225
  • Thank you received: 16
Ron,

Just to be clear, that image is high level zoom at some dim stars. The bigger stars just look bloated, which is what got me looking at the image in more detail.

I don't think it's collimation, as the circle would be lopsided... although it is a bit isn't it? I looked at my images from two nights ago and they had pretty good stars... so don't think it's the scope.

I'll check the lock knob and make sure it's loose... but pretty sure it's okay.

When I test tonight, I'll also try setting the step higher. Unfortunately, the seeing tonight is not supposed to be as good.

Thanks,

Ron
2 years 2 months ago #79814

Please Log in or Create an account to join the conversation.

  • Posts: 348
  • Thank you received: 69

Replied by Giles on topic For those with focus issues

Lack of collimation might throw the focusing out, maybe focus at bin x2, or collimate your scope for the future !!
2 years 2 months ago #79815

Please Log in or Create an account to join the conversation.

  • Posts: 225
  • Thank you received: 16
Giles,

Good thoughts!

Thanks!

Ron
2 years 2 months ago #79816

Please Log in or Create an account to join the conversation.

  • Posts: 1208
  • Thank you received: 559
Re Ron Debry's question about "Solution Pending" and also re @Ron Clanton's point about "selecting different stars in each iteration":

I made some changes to Linear focus in October 2021, so they would have been released as part of 3.5.6 and later that affected those two things.

Pending: The scheme prints "Pending" on the focus graph once it is in its 2nd pass and has found a position with an HFR that is "good enough to complete" (e.g. within the tolerance of the best HFR from the first pass). It is greedy, though, and keeps sampling positions/HFRs until it stops measuring lower/better HFR values. When it does stop seeing improvement, it will try again in that same position a few times. After a few failures to improve it stops, otherwise it keeps sampling and moving. In this 2nd pass it is using step_size/2, and once it passes the 1st pass' min position it uses step_size/4. Once you see the "Pending", it is guaranteed to stop and not fail. 

Selecting different stars. I too wondered if using different stars for HFR comparisons was adding noise to the scheme. Therefore, I made a change such that, when it compares HFR values from two focus images, it only uses stars from the same image coordinates in both images to make that comparison, if at all possible. Note, that the stars displayed on the the focus image, with HFR values reflect the star detections from StellarSolver, but may/may-not be used to compare the HFR values with a previous image.

You can find a more detailed description of the changes and pointers to the code here:
invent.kde.org/education/kstars/-/merge_requests/458
 
The following user(s) said Thank You: Giles
2 years 2 months ago #79817

Please Log in or Create an account to join the conversation.

Time to create page: 0.907 seconds