Question regarding frequent Autofocusing: Isn't it just wasting a lot of time?

Alan Brunelleandrea tasselliRuediger
26 replies764 views
Ferenc Szabo avatar
I do use an auto-focuser (Sesto Senso 2), but after the initial auto focus before any sequence, I don't. 
The reason is, because it takes forever. 
10 minutes or more sometimes.  It just keeps moving the focuser up to like 24 points until it finally settles on the best position (both in N.I.N.A and APT)

But people with filter wheel and have their set up to auto focus every hour (! ) or 1 degree change in temperature-  The night is short already during summer, not to mention clouds,  are you not losing like 30-50 minutes just auto focusing all the time? 

Here in North east, during summer months, the astronomical darkness is like between 11pm-3:30am and any hour of clear sky is very precious and rare due to unseasonably wet summer this year. 

There is no way I'm gonna have my telescope waste time to auto focus for 15 minutes after every hour or every time it drops 1 degree. That would make the 4hrs of darkness to be 2.

If it doesn't take long, how is your auto focus set up?
Helpful Insightful Respectful Engaging
Olaf Fritsche avatar
For me, one pass with the autofocuser takes about 5 minutes. For some filter changes, refocusing is unnecessary because both filters have almost the same focus point. But with other filters, the settings differ so much that I would have muddy stars without refocusing. Therefore, I accept the loss of time. 

I use the filters from ZWO. Would the focus position of different filters be better with more expensive suppliers? Or is it perhaps because the filters are not exactly parallel to each other in the filter wheel?
Well Written Insightful Respectful Concise Engaging
Sean van Drogen avatar
Have been contemplating the same and am going try and get filter offsets measured and use those to see what the results are. From your original post if you need 24 points to get a good focus curve it might be better to increase the step size. At the moment for broadband I use 5s exposures and 3 exposures per point to average out seeing conditions and usually usually need about 10 points so takes me 150s per filter. For narrowband I do the same but with 20s exposures which takes about 600s. The narrowband filter ones are the painful ones losing 10 minutes
Helpful
Alan Brunelle avatar
Edited: Meant to say Hi Frank!

Hi Olaf,

The time lost for autofocus can be a bit frustrating during these short nights!  But you get to control the way the autofocus does its routine (at least for the two sequencers I have used) and you should be able to put the settings so that you minimize the time.  But currently mine still takes about 6 minutes.  Some of the settings to look for is the amount of the offset used upon starting of the autofocus.  This is how far from the last in focus position do you want the autofocus to go to start the process.  This should not have to be too far.  This could be chosen by distance or by number of test increments from the last in focus point.  (See next point on increments.)  Then you should choose the number of increments (i.e. the size of the jump from one position to the next) for each new image acquisition in the process.  These should not be too fine.  Otherwise it will be a slow walk to get to focus, and then beyond, as is required of most algorithms to find best focus.  I also take two exposures at each test point so that the algorithm uses the average HFR value.  This can help if there are some "slight" errors in the images taken.  Some might do more, but then that adds a bit more time.  Not just the exposure time, but the calculation time as well.  

For my setup, I find a 2 second exposure is fine.  But I have a fast telescope.  But you might be able to fine tune the camera gain to keep your exposure times reasonable.  Upon starting, I ask the software to move only 4-5 steps away.  I also use 35 ticks of my motor focus (Celestron.  Every motor brand will be different, which you can work out as you learn about it when you focus manually with the motor.)  If you are shooting for in focus star HFRs of ~2.5, etc., then the software only needs to start outside of focus roughly at an HFR of 4-5.  Please understand these numbers are only examples per my setup.  Yours will very likely to be different.  But the bottom line is you should strive to get best focus in the routine with 10 or less data points.

Also, prior to starting an imaging session, it really helps the routine to get the scope as close to a decent focus as possible.  I usually have time to do a manually adjusted focus while aligning, etc.  And my motor seems to benefit from "exercising" it a bit anyway.  Often my first autofocus sucks, so I do one manually prior to a sequence start just to "proof" the system.  Related to this is what happens when your wheel moves to new filters.  I will let others speak to this, but again, one should strive to get the telescope close to best focus prior to starting autofocus.  Here there are settings to change that can accommodate non par focal matched filters.  The software should be set to move to a known offset position upon that change.  But others can speak to whether that is accurate enough for a 4 step focal offset to find focus most efficiently.

Alan
Helpful
John Hayes avatar
There is something wrong if it takes 10 minutes to do a V-curve!  The amount of time required between refocusing depends on the thermal sensitivity of the telescope and the local rate of temperature change.  I posted the results of a thermal sensitivity study here: https://www.cloudynights.com/topic/617666-thermal-sensitivity-of-casses-scts-and-the-c14-edge-hd/?hl=+thermal%20+sensitivity.  Although that study was centered on Cassegrain type systems, it discusses the issue in a larger sense and includes a chart of sensitivity values for various types of telescopes.  The thing about focusing is that yes, you may waste a little time by refocusing throughout a night, but if you lose focus and you don't check it, you might waste almost the whole night!

My suggestion is to take the time to figure out what's causing your system to take so long to focus.  I don't use V-curve focusing on my larger systems but when I'm out at DSW, I can hear all the scopes doing their V-curves and none of them spend more than about 3 minutes for a focusing run.  I personally use astigmatic focusing on my large scopes which runs constantly in real time and consumes zero imaging time, but that's not an option on many smaller scopes with limited back working distance.

John
Helpful Respectful Engaging Supportive
andrea tasselli avatar
I don't autofocus. Saves me bunch a £££ and a lot of time. And I don't even have a CF tube!
Alan Brunelle avatar
andrea tasselli:
I don't autofocus. Saves me bunch a £££ and a lot of time. And I don't even have a CF tube!

*Lucky you!  With my RASA 11, the response it yields was so flat that this coupled with the image shift (my old version RASA), made it hard for me to be sure I was in focus even when manually focusing.  So the motor was a god-send.  But the software was free, so making it auto was not that painful.  

I have to say that it holds focus pretty well, but for RASAs, John's table listed Hyperstars (no RASA info provided) as amongst the most sensitive scopes for temperature shifts.  Luckily I live on the other side of the Cascades from John, so we suffer less from those temperature changes.
Uwe Deutermann avatar
I believe most of it was said already. Auto focus is for me absolutely necessary, the temperature changes overnight are just too high. If I do not focus at least after 90 minutes it will be pretty bad out of focus. So definitely not a waste of time. The time that it takes for you is too high, too many points for sure, I use 9 at the most, AFTER I set the focus at the beginning of the night manually via Bahtinov mask. I am using SGP for autofocus, I think it could be better in determining the best focus spot, but at least it does the job fairly well. I do need around 10 minutes for the autofocus routine, that mainly due to the huge size of the 6200 images, the download sucks up the time. BUT … as you will read a lot from others as well: I'll take 10 better focused images rather than throwing half of them away. Yes, it is VERY frustrating to see the time passing by when the focus starts, but it is worth it!
Helpful Engaging Supportive
Roger Menard avatar
I use SGP and have it programmed to automatically perform autofocus when the focuser changes temperature by a prescribed amount. I use a 7 point autofocus and 8 second exposure. This only takes a couple of minutes. Most nights the autofocus routine is only activated a few times. I would hate to waste an evening of imaging with not-quite in-focus images.
Well Written Helpful Concise
Steven avatar
I've reduced the amount of autofocusing my rig does.
Yes, it has options to refocus every 1 degree difference, every hour, etc etc.

I usually find that temperature and time doesn't shift the focus too much. Not enough to really make refocussing needed.
Also, I shoot mono + filters in a filterwheel. And it does refocus after every filterchange, which usually is after 2-3 hours anyway.
The refocusing for the filters is "enough" for me. Also because I don't want it to "miss" the focus during a random-refocus.. and then lose out on images. 

On my colour rig, I focus once, and leave it. Now, I do have to say that, at the moment, the nights here are quite short. (realistically only about 4 hours of imaging time) And the nights also don't tend to have massive temperature changes. Once we get to the winter, and I can start shooting towards 8 hours again, then I might let it focus during the night 2-3 times to prevent any shift.

As for the time, no, it doesn't really take that long. Not 15 minutes at least.. something might be wrong there.
Usually it doesn't take me more then 3 minutes or so to go through the autofocusing (Using the ASIAIR)
Helpful Engaging Supportive
Blue avatar
I am using Voyager for capture and focus at least once per hour with my refractor, more often with my newt. 15 minutes for a focus run is a seriously long time, Voyager typically slews to a focus star, makes the focus run and slews back again in a couple of minutes max and it uses brighter stars for narrowband filters so they take about the same time as the LRGB.
Helpful
Roger Nichol avatar
I also use the Sesto Senso, it takes about 4 minutes to complete an auto focus on my Esprit 100ED. I could shorten that with fewer points but 4 minutes every 80 minutes is not a big overhead and produces sharper images.

Götz Golla avatar
I am using the ASIAIR and the ZWO autofocusser on a TAK 130D. I also found that autofocus takes many minutes, and also that the results are often not good. So why not focus manually ? With the ASIAIR, focussing manually from remote using the ASIAIR app on a mobile or tablet is possible and easy.  It usually takes me less than a minute with perfect results. Probably other setups require the observer to be at the telescope for manual focus, making it less convenient.
Concise
Ruediger avatar
Götz Golla:
and also that the results are often not good. So why not focus manually ?


Usually AF is very precise and on spot if the hardware is ok and the settings also carefully properly set. All professional instruments are focused automatically. 

Why not manually? Many points e.g.:
  • you work remote
  • you have a job and have to sleep during weekday‘s nights
  • There are scopes with no manual focusers at all
  • starting with a certain width you won’t get B masks, or they are hard to handle
  • manual focusing introduces vibrations
  • keeping the focus point during temp changes continuously perfect through all night
  • you are not interruption your imaging sequence
  • cold fingers in winter compared to sleeping well in warm bed
  • a.m.m.

This list can be extend by so many points…

CS
Rüdiger
John Hayes avatar
Roger Nichol:
I also use the Sesto Senso, it takes about 4 minutes to complete an auto focus on my Esprit 100ED. I could shorten that with fewer points but 4 minutes every 80 minutes is not a big overhead and produces sharper images.


Four minutes seems pretty slow.  You should be able to gather 10sx9 images in 90 seconds and the focuser shouldn’t take more than about 1-2 sec/image for another roughly 20 seconds.  Refocusing at the end should only take 3-5 seconds.  That’s a total of 115 seconds—call that two minutes.  Where is the hang up?  Is that particular focuser super slow running?  My feeling is that if you are spending more than about 3 minutes on a V-curve, there is room for improvement.

- John
Well Written Helpful Insightful Concise Engaging
Roger Menard avatar
As I said in a previous post, I use SGP. I have an Optec focuser. SGP tracks the results of the autofocus. SGP also includes an autofocus viewer which superimposes all of the autofocus runs for the night. By looking at the plot I can see that the focus does shift during the night as the temperature drops. While the change is not drastic, it is probably enough to make images taken later at night not quite as in-focus as earlier if I did not autofocus.
Well Written Helpful Insightful Concise
Götz Golla avatar
Ruediger:
Götz Golla:
and also that the results are often not good. So why not focus manually ?


Usually AF is very precise and on spot if the hardware is ok and the settings also carefully properly set. All professional instruments are focused automatically. 

Why not manually? Many points e.g.:
  • you work remote
  • you have a job and have to sleep during weekday‘s nights
  • There are scopes with no manual focusers at all
  • starting with a certain width you won’t get B masks, or they are hard to handle
  • manual focusing introduces vibrations
  • keeping the focus point during temp changes continuously perfect through all night
  • you are not interruption your imaging sequence
  • cold fingers in winter compared to sleeping well in warm bed
  • a.m.m.

This list can be extend by so many points…

CS
Rüdiger

Hello Rüdiger,
just a quick answer, because you misred my contribution: "With manual" focussing I did not mean turning a hardware focussing knob at your telescope focusser, but rather manually using the telescopes/focussers remote control software "knobs" from inside your cosy home,  and manually check  the result on the screen of your PC or tables/mobile  with a high frame rate of, e.g. 2 images per second. This is very quick and precise. I admit it doesnt solve the sleep problem, and it may not be available for some setups of focussers, network and software. It is easy and straightforward with, e.g., the ASIAIR, though.
Regards,
Götz
Helpful Respectful Concise
andrea tasselli avatar
Manually focusing images (and I mean REALLY manually focusing, no electric gizmos involved) is absolutely no bar in obtaining excellent images. I still don't see what is the problem with that, unless you only operate remotely. And if you are the kind that prefers sleep over starry skies then I suppose you are more keen in the P of AP rather than the A. Nothing wrong with that, mind you.
Alan Brunelle avatar
John Hayes:
You should be able to gather 10sx9 images in 90 seconds and the focuser shouldn’t take more than about 1-2 sec/image for another roughly 20 seconds.  Refocusing at the end should only take 3-5 seconds.  That’s a total of 115 seconds—call that two minutes.  Where is the hang up?  Is that particular focuser super slow running?  My feeling is that if you are spending more than about 3 minutes on a V-curve, there is room for improvement.


I have no reason to disagree with your reasoning here.  I would add, however, that cutting the focus sequence to the bare bones will risk the occasional rejected attempt and then suddenly your sitting through the time to do a repeat.  So adding an extra 2 data points can rescue a V curve if one or two points are off a bit for some reason.  Also, the focus response characteristic of a scope/focuser could play a part in choosing the settings, such as step number and step size.  My RASA with older style focuser is known to have a fairly flat response.  While I had my routine to use 4 steps on either side of focus, I had to move up to 5 the other night for some reason.  I suppose I could cut my step numbers down to 3 either side and dramatically increase my step size but I do see the occasional off point and I am more comfortable using more hard data points close to focus than An extended interpolation.  Worth the extra 90 seconds for me. 

I think an earlier post I stated my routine takes 6 minutes. But the data I looked up (7 min, actually) included a slew/plate solve.  The focusing took less than 4 minutes.
Helpful
Alan Brunelle avatar
andrea tasselli:
Manually focusing images (and I mean REALLY manually focusing, no electric gizmos involved) is absolutely no bar in obtaining excellent images. I still don't see what is the problem with that, unless you only operate remotely. And if you are the kind that prefers sleep over starry skies then I suppose you are more keen in the P of AP rather than the A. Nothing wrong with that, mind you.


I hear you!  But playing with the gear and software is kinda like playing with the focuser!  Actually, since my scope has no visual capabilities, I can care less about true manual focus. Manual is a pain with that scope. But I love doing So with my Mak Newt and when I get my new imaging Newtonian, I hope to fit it with a autofocuser that has a releasable clutch so that I can manually focus as well.  Both for AP and for visual use.

All that said, because I have no pier to work with and my tripod sits on a concrete block patio, I find that my results are always better when I work remote (about 100 ft!) Because when I work at the scope my HFRs are larger due to vibration and flexure under foot.

CS,
Alan
Bob Lockwood avatar
I have not heard anyone mention FocusMax, dose not anyone use this program? Is it free, NO $150.US But it works. Once it learns your system
(setup) and this is done once as long as you use the same setup. After that it simply dose the math, it takes a few quick images, dose the math and sets the new focus point. 1 minute or so to go through it’s initial focusing run, maybe 2 minutes if it needs to move to a brighter star and then a final check to reconfirm focus.
Ruediger avatar
andrea tasselli:
And if you are the kind that prefers sleep over starry skies then I suppose you are more keen in the P of AP rather than the A. Nothing wrong with that, mind you.


Sorry @andrea tasselli , but this is complete nonsense. This is a crude and in multiple logical ways false assumption:

1. If we would not be interested in the pictures (outcome), there would not be any need to focus at all (neither manual nor AF), since turning knobs is the aim of the game;

2. If we would not be interested in the image, we all would not have an account here and no images were uploaded here;

3. Vice versa! Since I/we love my/our equipment, I/we want to make use of it also during the week. Normal people have to work to ensure their lives and families and also earn money for our hobby. Moreover human beings need sleep. Hence you have to find automation solutions to make your hobby possible at all (BTW: Mine is AP and not turning knobs)

4. So in your opinion any owner, scientist or A using an AF is not a real A but an AP in your eyes? Wow! I think a lot of people think differently.

In my opinion: Everybody can do it in the way he wants to. However he or she has joy, fun and gets the results he/she wants to target for (pictures or turning knobs). But such assumptions are very disturbing (at least for me) and I am not accepting at all being categorized in any way, only by the fact using an AF or turning the knob of a focuser.

CS
Rüdiger
Alan Brunelle avatar
Bob Lockwood:
I have not heard anyone mention FocusMax, dose not anyone use this program? Is it free, NO $150.US But it works. Once it learns your system
(setup) and this is done once as long as you use the same setup. After that it simply dose the math, it takes a few quick images, dose the math and sets the new focus point. 1 minute or so to go through it’s initial focusing run, maybe 2 minutes if it needs to move to a brighter star and then a final check to reconfirm focus.

Hi Bob,

Thanks for pointing that out.  I took a look at the web page and as best as I can figure, it uses a different metric to measure the state of focus for a star during the V-curve generation.  The Half Flux Diameter (HFD) is claimed by them to be a better measure for a wider variety of instruments and over a wider range of difficult conditions.  So that is interesting.  I do not have the time to read the full paper just now, but do you have anything to add to what I wrote here?  I am not sure I would actually see a difference with my setup.  Nor would the potential improvement, if attainable with my setup justify the cost, however.  Unless you can add more info... 

What I am concerned about is the FocusMax may indeed provide a better accuracy for state of focus at points that are well outside of perfect focus because this is where they state that it outperforms.  But given enough points close to perfect focus, I am not sure that HFD can outperform HWHM/HFR, since there is not a significant difference in the PSF close to, but not at, perfect focus.  So maybe the V-curve is straighter with better extreme points inside and outside, but if the curves are symmetrical inside and outside, each method may still arrive at the same best focus.

Thanks and CS,
Alan
Helpful Insightful Respectful Engaging
andrea tasselli avatar
Ruediger:
Sorry @andrea tasselli , but this is complete nonsense. This is a crude and in multiple logical ways false assumption:

1. If we would not be interested in the pictures (outcome), there would not be any need to focus at all (neither manual nor AF), since turning knobs is the aim of the game;

2. If we would not be interested in the image, we all would not have an account here and no images were uploaded here;

3. Vice versa! Since I/we love my/our equipment, I/we want to make use of it also during the week. Normal people have to work to ensure their lives and families and also earn money for our hobby. Moreover human beings need sleep. Hence you have to find automation solutions to make your hobby possible at all (BTW: Mine is AP and not turning knobs)

4. So in your opinion any owner, scientist or A using an AF is not a real A but an AP in your eyes? Wow! I think a lot of people think differently.

In my opinion: Everybody can do it in the way he wants to. How ever he has joy, fun and gets the results he wants to target for (pictures or turning knobs). But such assumptions are very disturbing (at least for me) and I am not accepting at all being categorized in any way, only by the fact using an AF or turning the knob of a focuser.


I guess I didn't express myself clearly on this subject. Just to clarify, A stands for Astro and P stands for Photography. No more, no less. If your aim is just for the nice pictures then it is just another form of landscape photography (from which many come to AP, by the way). If your aim is being under the stars and ALSO take pictures of the night sky then you are likely more keen on the astronomical part of it, as an amateur obviously. Personally, I don't give a damn about the equipment I use, so much so that I rent time when I feel the need. But that does not supply the needed fix of being out in the night. And this brings in the matter of auto-focusing as a necessity, which is not. If you are out there in the night then refocusing is just one of the many things you might enjoy about being out there and taking pictures, vicariously connecting with the visible universe via lens/telescopes/camera AND making an effort for it.

N.B.: All this refers to your earlier post about the many reasons of AF vs. manual focusing. The only one I can agree with is the first one.

Besides, we all work for a living and me no less than any other. So, yes, short sleeping night and long waking days but that is the lot I enjoy. Anyways, to each his/her own.
Ruediger avatar
I am afraid, but I completely disagree. Turning a knob does not decide anything at all and by far not whether you are an A an AP.
Related discussions
Autofocusing from a Remote Site
Hello, Does NINA have a log of error messages? I ran autofocus, stars look fine, but at the very end I got the message saying autofocus couldn't complete. It seems to have completed. What should I do when I get this message? I'm using the Ses...
Aug 6, 2024
Both posts discuss frustrations with the slow performance of autofocus routines when using the Sesto Senso 2 focuser with astronomy software.