Official Request for Improved Autofocus in SGP


#61

The SGP auto focus routine already supports backlash compensation. It works quite well. You just need to specify how much to use, which is fairly easy to figure out. Larger amounts than required do not cause a problem, so it is not a sensitive parameter.


#62

Hi,

A robust autofocus strategy is immune to backlash. If you always walk one slope of the focusing curve to determine the focus point, backlash compensation is not needed at all, the price to pay is that you have to build a focusing model a priori.

Cheers,

Jose


#63

It does not do any harm to use more backlash compensation than necessary. It’s okay to overshoot backlash compensation. If excessive backlash compensation is not working, then there’s something wrong with the mechanical focuser hardware and/or motor.

Peter


#64

There is really probably not enough juice in the orange to expect much on the focusing issue any time soon. The fact that this has been on the agenda for so long and that not much was done beyond the prior effort to improve focus is evidence. Same with multiple instances and multiple cameras. There is a certain amount of denial of the issues that people are having for various things. There was hope that for the last pay increase there would be some progress and there probably has been some progress, but I would have personally stayed with the 2.6xx versions.

However, even with a pay increase, there is still not enough juice in the orange to generate a large effort at recoding things. I suggest that people punt and integrate FocusMax into the equation. That won’t fix other issues that are listed in the forums, but it will hopefully be easier than to continue to work on a proprietary model for focusing. Then the other issues can be addressed as time allows. Focus is mission critical and for those who think that the focus issues are associated with people who just don’t follow directions, or don’t know what they are doing, or have bad equipment. You are just outright wrong. I support integrating FocusMax and not re-inventing the wheel. Then maybe people will have a proven method of focusing larger scopes with central obstructions or faster scopes with central obstructions.


#65

I would add that in the scenario I indicated in my prior post, if people don’t want to pay anything more for focus max, then just stay with the SGP proprietary model and leave it as is. Those that are happy with should just leave it alone.


#66

One additional comment. I am not complaining about the price increase, instead I am saying if you go to the well, go once and get a full bucket of water to quench the thirst that arises out of hard work and creates the incentive to continue.


#67

I look at it in a slightly different way. FocusMax commands a fairly serious charge relative to the cost of SGP and only gives us one function whereas SGP gives us much, much more. I don’t have the experience of seeing how well it works, but if the development effort is roughly in line with the price, FM must do some very clever and complex calculations to work out the ideal focus. Therefore, is it reasonable to expect SGP to replicate the research and develop the code with just some good suggestions from the user community. I appreciate that it has been an issue for some time, but developing the code can only come once the mathematics behind what has to be achieved is fully understood. The code should then be easy. Can anyone provide the fine detail behind how to achieve focus using faster, more reliable methods like those used by FM ? Sorry CCDWare, but we all want what you provide included in the SGP price. I run an SCT and lose about 10% of my subs due to focus issues, but I still can’t see that I can justify the cost of FM.


#68

Great, so for us with zero backlash focusers, this idea would do nothing.


#69

I don’t understand your point Bill. I either use the backlash compensation feature in SGP or in my module’s firmware. It works fine with both Crayford and R&P mechanisms. Since I use the same module to control different telescopes, I just plugged in a backlash amount that slightly exceeded the worse of the motors and I never give it a second thought.


#70

I’m guessing the focus process boils into several sections which we are all discussing interchangeably:

  • Strategy (single/multiple, full V-curve, sampling and so on)
  • HFR calculation
  • Motor control
  • Recovery modes (outliers, multiple sampling, out of range)

I have several refractors and one RCT. I generally do not have any operational issues with the refractors (f/3.5-F/8) but the RCT needs more care. I used FM several years ago but never in combination with the RCT but I do occasionally use TSX when I’m doing a T-Point run. I have noticed that TSX is better able to determine the diameter of an out of focus star (donut) than SGP, which sometimes selects hotspots on the annulus (due to seeing).

I have also noted that if my RCT is not perfectly collimated, the V curve is wonky, with different slopes. It is actually a really good indicator of collimation ;).
There is nothing that SGP or any other program can do about that and if an algorithm were to rely on one side of the slope, it would get it wrong and the user would be unaware. In this instance, so long that I do not go too out of focus on the AF run, SGP at least still finds the best position.


#71

I’m a very happy user of SGPro for about 8 months and would like to add my support to having a more robust Autofocus routine.

Thanks and CS,
Nick


#72

There are other focusers on the market that dont use R&P or Crayford technology. To decide on a method based on those would be limiting, and again is completely unnecessary.

The myopic perspectives here are all interesting to hear from a problem perspective, but are not all that useful from a solution perspective. You can take all of these variables out, and come up with a solution that can work across camera types and focuser types and scopes with or without CO.

I feel the need to make it very clear (since I was the vocal person on “the other site”) that I do think that SGP is a good Astro Imaging platform. While I dont subscribe to the notion that I should just “accept what it is” and honestly dont really understand why others would (after all, you paid money for this), I do think that the focusing method in this software is a bit long in the tooth.

I hope this aspect of the software can improve.


#73

Good point - there are others too, but by definition anything mechanical and with moving parts has a tolerance and hence backlash of some order. The ‘method’ ultimately looks at the optical effects of in and out of focus stars and the backlash aspect only affects the ‘motor’ control aspect of the process.

I think the word ‘myopic’ is somewhat inflammatory, maybe parochial is more accurate. The problem is that we all have different equipment and different experiences, which we are sharing here. I respect what the inventors of FM did for the hobby but I do not miss the old free version of FM. At the same time I can also see room for improvement in TSX and SGP focus algorithms too. TSX have recently overhauled theirs. I have only used it a few times and my early assessment is that it gets there (I think) after quite a bit of time but I have not yet really worked out what it is doing, as for me, it does not have the visual feedback that a well graphed V curve has.


www.mainsequencesoftware.com