Hello everyone.
With this post I would like to have knowledge of a factor about AF.
I have been a SGP user for many years, with 2 different setups.
I found that with a setup I made an improvement with the original focuser, in an Esprit100 I put a 3" Esatto (due to weight and tilt problems in the original) now everything is perfect.
The thing is that I notice that my automatic focuses are just as slow as before I used a Sesto Sento. There is no need to go to a Ha filter, a Lum filter and bin2 are enough.
I use 9 steps in bortle 3-4, I have good sky, good seeing, and I get the feeling that the routine is very long. A friend uses Voyager, with the same setup as mine the AF is very fast.
My question is, is it a matter of my configuration? Or is the SGP AF motor slow? With my other setup (Samyang lens, ZWO focuser) it is just as slow. I would appreciate a little more wisdom on this topic.
Thank you so much.
Are you saying that you are running AF for narrowband filters using the actual narrowband filters? Or maybe I am misunderstanding.
If so, you can absolutely focus for narrowband filters using broadband filters and doing so will yield a considerable time savings.
Maybe I explained myself wrong.
I say that normally a narrow band filter takes longer to do the AF, the problem I have is with broadband filters, or narrow band, the type of filters does not matter.
If I make comparisons between narrowband or wideband filters, the slowness problem is the same.
I don’t know if I explained myself well now.
@dracolino, you mentioned SGP AF motor slow. SGP only tells the focuser’s motor where to move to and has no control over how long it takes the motor to move there. Other than the initial move to a higher step count which can take several seconds, the moves from one AF image to another are much shorter. Are you saying your AF motor is slow? If so, how slow is it? For example, run a manual AF to find the focus point, then run another one and time how long it takes the motor to move the first time, i.e., from the focus position to the highest position. And time how long the whole focus run takes.
When using narrowband filters, most people use the Luminance filter to focus and have filter offsets for the narrowband filters. This significantly decreases the time to focus.
The time it takes to focus depends on:
- How long it takes the motor to move - this is usually a small percent of the total time.
- How long your exposures are.
- How long it takes the camera to download an image. CCD cameras can take up to 15 seconds.
- How long your computer takes to process an image.
- How many exposures are taken.
- How close to focus you were at the start of the focus run. If way off focus, the focus routine will likey start over once or twice, which doubles or triples the total time.
Can you upload a log file and tell us when the slow focus happened?
Thanks for your answer.
I understand that SGP tells only the focuser what to move, but I suppose that the “power” of SGP will make its calculations faster or slower. Correct me if so.
Now that you mention doing a manual focus and then doing it automatically, many times the manual focus is faster than the auto focus, and if then I start the sequence and the auto focus starts as you say, many times it has to do the focus twice, with “star size” errors
Using compensations has never been useful for me, I prefer to focus with each filter always, for me safer, call me conservative, or manic, haha.
To your questions:
1.The time it takes for the motor to move is usually a small percentage of the total time.
- I’ll calculate exactly tonight. But I work remotely.
2.How long do your exhibitions last?
-Approximately 2 to 3 seconds. Narrow band 6-7sec.
3.How long does it take for the camera to download an image. CCD cameras can take up to 15 seconds.
-It is a QHY268M, it usually takes about 5-7 seconds.
4.How long it takes your computer to process an image.
-About 5 seconds approx.
5.How many exposures are taken?
-9 is what is configured.
6.How close you were to focus at the start of the focus stroke.
-Think that the travel of my focuser is only 2.5cm, normally there is not a very big difference from the previous day.
Following the formula to calculate the steps, I would have to set 3500 steps, but my tests do not work with these steps, they have to be 7500 to 10,000 steps.
I will calculate an automatic focus with a clock, but from memory, I would say 4 minutes, you can do it.
Let me send the log tonight with my evidence.
A comment: In the latest release of SGP that I am using I have noticed that the SGP evaluation time of the autofocus frame has become about twice as long as in earlier releases. Nothing else has changed in my setup, so I take it that something has been modified e.g. in the calculation of HFW values and how they are used.
We are unable to produce any results like that here. What size chip are you using? Did you enable eccentricity measurements?
I have deduced that, that perhaps the eccentricity measurement is the cause, could it be?
@Ken This was not meant to be a bug report but rather an observation.
Autofocus comprises multiple steps: Setting the focuser position, capturing an image, downloading the image, analyzing the image, and showing a new point on the autofocus curve. When I was running autofocus the other night the time analyzing the image before the measured point occurred on the curve was about twice as long as in my runs some week ago. The captured image window also blacked out for a second before the point was displayed on the curve. I noticed that the captured window image had all the HFR-values displayed, which I am unsure happened before.
I am not aware of any eccentricity measurements and I can not find any setting to enable or disable them.
I am currently running SGP version 4.3.0.1331 on Windows 10 version 19045.4291. The PC has an Intel(R) Core™ i7-6820HQ CPU @ 2.70GHz processor with some 8 GB RAM, so I do not think PC performance is any problem.
Maybe a clarification on those eccentricty mreasurements could help clear things up a bit?
All the best,
Rolf J
Yes, depending on your machine and image sensor size, eccentricity measurements can have a huge impact on the time to analyze an image. As such, we added an option to enable or disable it. If eccentricity measurements are not important to your workflow you shouldn’t have to incur the time penalty.
I’ll add this to the helpful and as a tooltip on the option’s checkbox.
I has sworn that I updated the help files with this info, but, upon reviewing them, I don’t see any mention. In any case, you can find this option using Tools > Options:
The truth is that it is not necessary for me at this moment, only when I make camera changes, backfocus, etc… I will try to remove this option and I will tell you the results.
It would be nice to be able to implement an option in the AF, and that is to calculate the time it took to do the AF.
Thank you