×

INDI Library v1.9.9 Released (30 Nov 2022)

Bi-monthly INDI Library released with new drivers and bug fixes.

Ekos Autofocus

  • Posts: 10
  • Thank you received: 1

Ekos Autofocus was created by Todd Boland

Hi all. I am wondering what your autofocus plots looks like. I am new to this hobby and for all I know, this could be normal. I was expecting more of a V shape based on what I have read about autofocusing. When I run autofocusing, I usually see a mash of very similar measurements. I would imagine a small step size and starting out close to the optimal focus will result in plots like mine but I'm not sure. Can anyone comment?

The attachment is from tonight using a RASA 8 with a ASI533mc. I am using a ZWO EAF driving a feather touch. I have had very similar experience using deepskydad v3 with a RedCat so I suspect this is normal.

2 years 5 months ago #58410
Attachments:

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

  • Posts: 930
  • Thank you received: 432

Replied by Hy Murveit on topic Ekos Autofocus

No, I would not be satisfied with v-curves that look like that.

As an example, I'm attaching images of several v-curves from my session the other night.






Admittedly, I'm using a 105mm refractor (with a Moonlight V2 controller), which I'll bet is much easier to focus that a RASA, but you should aim for V-curves.

I'd recommend
- use the Linear Focus algorithm
- use the SEP detection
- use Full Field with a 20% 80% annulus
- 5% tolerance
- no dark frame
- I expose for 3-5 seconds using my f/5.6 scope.
- You'll need to figure out your "initial step size".
If the initial step size is too small, then you might get noise, like your plot shows.

Take some time to get things right the first time. So, under decent skies, first manually focus, until you have an image
that's sharp to your eyes, or better yet, get sharp focus manually using a Bahtinov mask.

Once in focus, tune your initial step size by running autofocus as I described above. Play with the step size.
Here's how you might get an inital step-size value: move slightly off focus and find a step size that moves the focus from an HFR of 1.5 to 1.25 or something like that.
Not too important to be exactly that, but you want to see some change in the focus HFR.
[The focus tab has a "start framing" button, two arrows pointed opposite ways, that can help with doing this].
Now that you have an initial guess for the step size, run autofocus. Hopefully you'll see a decent v-curve.
If you want, change the initial step size a little and iterate. Best to get a decent curve, even it you wind up with 15-20 samples,
than something quick that doesn't look reliable.

I'd avoid playing with max travel (keep it very large), backlash (keep it 0), out-step multiple (keep it at its default of 5).

Let us know how it goes!
Hy
AP1100 & Orion Atlas Pro, WO/ZS105 w/Moonlight V2 focus, GSO RC10 w/RSF focus
ZWO ASI1600, Astronomik Filters, ST80, QHY 5L-IIm.
KStars/Ekos/Indi on NUC10 & RPi4 w/SSD -- Ubuntu
Projects: Greedy Scheduler, Terrain, Polar Align, Analyze, Linear Focuser, SEP MultiStar & GPG Guide, FITS autostretch.
The following user(s) said Thank You: Brian, jiberjaber
2 years 5 months ago #58420
Attachments:

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

  • Posts: 412
  • Thank you received: 30

Replied by Paul on topic Ekos Autofocus

Astrodd?,

Would you please post your focus settings, I also use a feathertouch focuser and may be able to suggest some setting changes.

Paul
Losmandy G11 with Gemini 2 controller
QSI 583/683 monochrome cameras with filter wheel
Starlight Xpress Lodestar X2 guide camera
Microtouch Focuser controlled using Arduino UNO emulating Feathertouch controller
Raspberry Pi 2GB with Raspberry Pi OS 64bit Bullseye
2 years 5 months ago #58424

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

  • Posts: 391
  • Thank you received: 116

Replied by Doug S on topic Ekos Autofocus

@astrodd: I have a RASA 11, and ZWO EAF focuser. Our setups are sufficiently similar that I think I can help. Hy's comments are appropriate, but I would add a couple of items. Do NOT try to use the Iterative or Poly algorithms under the process tab; they just won't work for RASA + EAF. Do use Linear, and adjust your "Mechanics" tab settings. Until you understand your RASA 8 temperature & elevation focus sensitivity, you may need to leave your Out Step Multiple to 5 or higher, but once you get one good focus, you can reduce it to 2.5 or 3. This will shorten up the autofocus run time dramatically. Use an initial step size of 200. Note that the 1st generation RASA focus mechanics have some internal instability that may manifest as an upward bump on the right side of the V curve approaching focus. If you see what looks like a 3rd order poly function, set your initial step size a bit bigger (250 or 300). This will likely jump over that region and produce a nice quadratic (V).

Finally, I strongly recommend you use the autofocus log to graph your focuser position vs temperature. This will allow you to "seed" your focuser position at the start of the night (or anytime you walk away for a while) so that the linear algorithm will work more efficiently. Otherwise, you'll need to widen your Out Step Multiple, causing long autofocus run times. FWIW, using the autofocus log and graphed results, my RASA11 + EAF focuser temperature sensitivity is shown below. Note that the spread (above/below trend line) is mostly elevation dependency....a work in progress to improve! I would expect your offset and response to be a bit different (8 vs 11), but not radically. Cheers, Doug

RASA11, Celestron CGX-L, ASI183mc Pro, 60mm guider + ASI290mm mini, ASI EAF focuser, PPB, Rpi4-4Gb+SSD, Powered USB3 hub, hardwire Ethernet.
Last edit: 2 years 5 months ago by Doug S.
2 years 5 months ago #58436
Attachments:

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

  • Posts: 10
  • Thank you received: 1

Replied by Todd Boland on topic Ekos Autofocus

Wow, I sure do not regret asking this. There is so much information packed into the replies. For the record, I'm using the default settings. I only tweaked the settle value.

I plan to set up again tonight and try the advice offered. Thanks Hy and Doug.

I will also start a temperature vs position log. I can see how this would help seed the focus. This is a brand new RASA 8 from OPT so I suspect it doesn't suffer from the focus issues from the first batch.

I hope my gratitude shows through. I genuinely appreciate your replies.
2 years 5 months ago #58453

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

  • Posts: 190
  • Thank you received: 28

Replied by Sonny Cavazos on topic Ekos Autofocus

I have a moonlite focuser and I use a backlash of 75 or so. I think you are seeing a lot of backlash on the focuser, my nitecrawler works like the nice graphs HY posted.

On my moonlite, I also tightened the two set screws as my focus drawtube was a little lose.

But after backlash setting, my focus works a lot better.

I also make sure I focus with a camera offset of 0 as it helps in my case.

Thanks.
Last edit: 2 years 5 months ago by Sonny Cavazos.
2 years 5 months ago #58454

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

  • Posts: 522
  • Thank you received: 103

Replied by Jim on topic Ekos Autofocus

I have a moonlite focuser, though on a refractor, not on a RASA. I have noticed that IF the focus is far off to start with or for some reason the algorithm steps too far beyond focus, then it gets lost and the graph levels off. That is, on the focus-in side of the V, if it goes beyond the high point on the left side of the V, it won't return, and no matter how large the stars get out of focus, it senses them as being about the same FWHM after that. I have only had this happen a couple times, it gets to the step limit, aborts, resets, and then works the next time. If you did not already, try to focus manually until it looks good, then start the routine to see if this makes a difference.

My settings are:
linear
SEP
2% tolerance
full field
20%/70%
initial step 40
multiplier 4
12" pier with HDX110 using EQMod
ASI 1600 mono/color mains with ASI290MM in off-axis
ASI filter wheel
Moonlite focusers for the sharpening
AT115EDT w/.8x for the light
Fedora Linux, 100% INDI
2 years 5 months ago #58456

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

  • Posts: 930
  • Thank you received: 432

Replied by Hy Murveit on topic Ekos Autofocus

Jim,

If you want, send me some images (e.g. attach a link to a drive folder) for a series of full-field images of out-of-focus to in focus, in regular steps (e.g. at the step-size you're using), where it starts from well beyond the point where you say the HFR measurements "flatten out" so that the Linear Focus algorithm can't deal with them. I can look and see if there's something to be done.

Linear focus itself is totally reliant on the HFR measurements, so if the measurements are bad, there's not much Linear Focus can do.

The HFR is computed by the SEP library, and it's likely that that software was not developed to deal with way out-of-focus images, but rather measure HFRs for relatively in-focus stars.

Hy
AP1100 & Orion Atlas Pro, WO/ZS105 w/Moonlight V2 focus, GSO RC10 w/RSF focus
ZWO ASI1600, Astronomik Filters, ST80, QHY 5L-IIm.
KStars/Ekos/Indi on NUC10 & RPi4 w/SSD -- Ubuntu
Projects: Greedy Scheduler, Terrain, Polar Align, Analyze, Linear Focuser, SEP MultiStar & GPG Guide, FITS autostretch.
The following user(s) said Thank You: Jim
2 years 5 months ago #58476

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

  • Posts: 379
  • Thank you received: 57

Replied by Jerry Black on topic Ekos Autofocus

[EDIT] I've tried to make the image visible and included a log file

I don't want to hijack this topic too much, but I'm certainly not able to use my Pegasus FocusCube with the parameters I have tried so far.

The last attempt used these: Full Field, Annulus 20-80%, Settle 2 sec, SEP, Linear, 5% Tolerance, Average over 1 frame, Kernel Size 5, Sigma 1.5, Initial Step size 200, Max Travel 10000, Backlash 15, Settle 1.5 sec, Out Step Multiplier 1.5.

The previous manual Bhatinov Mask focus gave a value of 62940 which is near the minimum observed for focus point #6 in the graph below.



I'm using a Triad Ultra Quad Filter with a Skywatcher Esprit 120 mm with a Nikon Z7 camera. It seems at the extreme focus points which show low HFR values, that what were single stars are being captured as multiple smaller points of light (i.e. smaller better focused stars than the larger out of focus star that should be reporting a higher value).

Any suggestions would be much appreciated. :)

The focus begins at 2020-08-20T21:43:01.151 ADT

File Attachment:

File Name: log_21-27-38.txt
File Size:2,409 KB
120 MM Skywatcher Esprit + Celestron CGX, ZWO ASI 224MC guiding, Pegasus FocusCube+PPBA
Nikon Z7 8256 x 5504 4.34 um. Triad Ultra Quad NB Filter
1) MeLE Quieter3C 8G Ubuntu-mate 22.04.1 2) Odroid-N2. 3) StellarMate RPI4 8G
The good, the bad, the ugly at sciencedowneast.no-ip.ca/zenphoto/
Last edit: 2 years 5 months ago by Jerry Black. Reason: Attempting to make the attachment visible and include a log.
2 years 5 months ago #58481
Attachments:

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

  • Posts: 930
  • Thank you received: 432

Replied by Hy Murveit on topic Ekos Autofocus

Jerry,

Can't see your picture (it says "You do not have permissions to access this page.").
Why not just attach it to the forum post?

I guess the questions to ask are:
- Are you starting autofocus reasonably close to an in-focus setting?
- Is your step size appropriate?
If you step size is too small, you just get noise on the measurements.
If your step size is too large, Linear could be going too far out when it starts its focus sweep.
Feel free to point me to a log, if you have verbose focus logging on.

Hy
AP1100 & Orion Atlas Pro, WO/ZS105 w/Moonlight V2 focus, GSO RC10 w/RSF focus
ZWO ASI1600, Astronomik Filters, ST80, QHY 5L-IIm.
KStars/Ekos/Indi on NUC10 & RPi4 w/SSD -- Ubuntu
Projects: Greedy Scheduler, Terrain, Polar Align, Analyze, Linear Focuser, SEP MultiStar & GPG Guide, FITS autostretch.
2 years 5 months ago #58482

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

  • Posts: 114
  • Thank you received: 2

Replied by Joshua R on topic Ekos Autofocus

Hi,

Is there a HFD value at which to say now we are in focus?
Mount: AZ EQ 6GT
Telescope: Newton 10' f4, Esprit 100ED, Celestron 8SE
Camera: Zwo 120MC, 294 MC Pro
2 years 5 months ago #58485

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

  • Posts: 348
  • Thank you received: 69

Replied by Giles on topic Ekos Autofocus

No, it is subjective, your lenses will only achieve so much focus and the value will be different depending on your equipment and the image you are trying to focus.

You should just look for being at the bottom of a smooth curve for a particular set of equipment and image target.
__
I blog my progress: www.coochey.net
2 years 5 months ago #58488

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

Moderators: Radek Kaczorek
Time to create page: 2.570 seconds