I've been having a bit of trouble with auto select star for focusing and guiding. About 35% of the time the guiding and/or focusing modules miss and select background with no stars. In these cases the focusing or calibration will be attempted regardless of no stars being present. When this happens I can abort and restart the process, and it usually finds a star when I do that. I often have to click "reset" on the focus tab to force it to get rid of the subframe (maybe always, not sure). Exposure length doesn't appear to be a problem as there are plenty of stars with good signal in these cases. Guider is set to 10s binned 2x2, and focusing is on L at 3 seconds, 1x1.

Any ideas on how to improve the success rate? I see mention of moving to sextractor for star identification. Maybe this is what is needed?

yeah I'm working on that. but 35% is too low.. in my case, I'd say 99% it gets it correct. Do you have a FITS image of it failing to auto detect a star?

Yeah, I thought the failure rate seemed kinda high. It always seems to happen when I'm not looking.

I have a guiding subframe that caused the failure last night still in the guide module, but when I open it in the fits viewer and save it, no other program (pixinsight, kstars, tsx) can open it. They all give an error. You can see the file here:

www.dropbox.com/s/l739tx3hpwzlhmy/nostars.fits?dl=0

Here's a screen capture of it. Those are hot pixels, not stars. I've see it have a star in the subframe and miss selecting it too. The focus module sometimes selects multiple close stars as a single star. A good example is the trapezium in M42.

I just fixed the fits problem when saving, the file was not flushed yet until you close it. Will check regarding using sextractor for finding stars.

Thanks. I've rebuilt and will try again.

In the meanwhile I turned "Detect Stars in Image" on the guiding tab with the image I made the screenshot from above with. This is the result:



I see similar results with the focus module when it happens. There are two different cameras at play here. The main camera is an FLI ML16200, guider is an SX Ultrastar.

Happened again tonight. Here are some fits:

www.dropbox.com/sh/e705ujqttflnw2a/AADPE...t_a9Jyb9gEPxrna?dl=0

There's a full frame as captured by the focus module, and a sub that it picked with no stars. Finally, restarting it had it find a star. A sub of that found star is included. Focus completed at HRF 1.34.

I just opened first file and 100 stars were detected! So something off is going on. Do you have the log? You need to enable "Focus" in the logging as well. At any rate, I'm working on the sextractor stuff, but I think this is a different problem as stars are getting detected but I guess something is off with the subframing.

Can you actually test this? In capture module, capture a binned image, then try to specify a subframe and capture.. is it the correct subframe? Try unbinned as well. Make sure CCD + INDI + Focus logging is ON.

Hi!

I have something that sound precisely the same. Here is a log file where the focuser fails to identify stars, instead focusing on noise. And a fits image here:

www.dropbox.com/sh/fs744ujp8oed79d/AACfn...E98N8km8Tv2dUBa?dl=0

Edit: No this seems to be a bad example. Totally out of focus I think.

Magnus

I'll give those tests a shot once the weather clears up. In the meanwhile, I've attached the focus log that goes with the images above. I didn't trim it. The first focusing attempt is the one that resulted in no stars. The second is after I reset it. Some observations:

- The subframes are close to each other but do not overlap.
- If I use pixinsight and create a preview for each subframe as defined in the log file on the full field, one of them has a star dead center and the other is just background. This at least confirms that that part of the image has no stars.
- The coordinate difference between these two subframes is +75,+51. This made me think of this post I made a few days ago. Although the axis are swapped, the numbers are identical. Maybe not related, but very interesting.

Thanks, I need to know if there is a problem in subframing in FLI driver (unbinned and binned). Maybe you can test it offline? If you can use some lens or identify some part of the image, subframe to it and see if you get the correct result.

If we don't need stars I guess dust donuts will do. I pointed at my flat panel and let ambient light illuminate it. Results....

NOTE: I don't use binning for focusing or alignment, so I'm not sure if this is relevant. I only use binning on my guide camera (SX Ultrastar). I did not test that one yet.

I set up an image capture as follows:
- Light001 binned 2x2 full frame
- Light002 binned 2x2, subframe set to coords 1790,1448 with 1000x1000 size <- did not subframe!
- Light003 binned 1x1, subframe as above <- did not subframe!
- Light004 binned 2x2, subframe as above <- subframed, 500x500 in size, right location
- Light004 binned 1x1, subframe set as above <- subframed, 1000x1000 in size, right location

I presume that the subframe coordinates and size are absolute for the sensor and not the apparent pixel size when binned. Please confirm.

Fits files, log files, and the .esq file are here:

www.dropbox.com/sh/p0zgs1vl5td91hn/AAAS5...IzQQGIbCWmKBZZa?dl=0

EDIT: I ran this for the guider as well, everything is in the dropbox folder above. The guider behaved normally once I got it to work. Seems the guider only likes origin coordinates that are divisible by 3 when binned 2x2? Took me an hour to figure this out. Hmm...

Ok I have some idea about what is causing this and just send you chat invitation in Google chat so it would be faster to solve it there.