×

INDI Library v2.0.6 is Released (02 Feb 2024)

Bi-monthly release with minor bug fixes and improvements

New Polar Alignment (Plate Solve) not working

  • Posts: 40
  • Thank you received: 3
For me I get an error that the CCD can't start a new capture because it is already capturing. I am using a ASI120mm mini to do the alignment, and so in most cases it captures in 1 second and solves in 1 or so seconds, meaning that a refresh rate of 8 seconds should be more than enough to capture and solve the images, but it simply errors completely.

I know the CCD and solve can happen properly because the 3 images needed to figure out the polar alignment error work just fine and solve quickly. Something in how this looping process is done is not working. I suspect it is timer based. Really, the time input should be the time between complete exposure + solves, and used just to prevent rapid updates or to let users time their adjustments.
1 year 6 months ago #86134

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

  • Posts: 1208
  • Thank you received: 559
FWIW, the label refresh rate is misleading. It really means the exposure time. That is, if you set refresh rate for 8s, then it will expose for 8s and then try to solve, and when that is all completed it starts the next exposure immediately. So if it took 1s for the solving process, you'd see a new exposure start every 8+1=9 seconds.

I assume this is related to the CPU being used, or StellarSolver settings, or something like that. I have not been able to reproduce this issue, though I have been looking into it. I do get occasional failures to solve, but I get solutions 90% of the tries. If someone could reproduce with the simulator and show me detailed "hot to reproduce" including logs and settings, then I might have more luck in debugging this.

Logs were disabled in StellarSolver because that was one of two things that seemed to cause random crashes in stellarsolver. Without the logging and the "partitioning", StellarSolver has been much more stable, IMHO.

Also, of course, you can always use the Move Star methods if this fails for you. Naturally, I'd love to fix it. Solving issues seem to always be tricky.
Hy

PS Is everyone experiencing this issue using a Raspberry Pi? Is it pointed at the pole (where there are few good alignment stars) or elsewhere in the sky? Can you try pointing the telescope somewhere else and see if the issue persists?
1 year 6 months ago #86138

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

  • Posts: 333
  • Thank you received: 23
I use a RPi 4 with 4GB of RAM, but last time I used 1sec for the refresh that is enought to see the star used as reference for the alignment procedure.
Last time i was pointed at about 45 degree WEST from the Polaris'field.
As you wrote, to use the Platesolve function and have the time to do the solving procedure I have to increase the refresh time to at least 4seconds
1 year 6 months ago #86141

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

  • Posts: 1309
  • Thank you received: 226
I've also experienced it becoming unresponsive after a few frames but I'm not sure if I'm using the most up to date version and need to test it more thoroughly before i can be more specific. And I am running it on a Pi under Raspberry OS.

Please consider making the refresh frequency into a timed delay between exposures under the solving method.

Thank you
1 year 6 months ago #86142

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

  • Posts: 42
  • Thank you received: 2

It isn't what's happening though. It just fails immediately after exposure.

I'll see if I can get you some logs.

I am using an RPi4 with 8GB RAM I own a CEM26 with iPolar, which is how I do the alignment. Typically the plate solver takes less than a second to plate-solve the iPolar's image. Yes, it is pointed at the NCP.
1 year 6 months ago #86143

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

  • Posts: 1208
  • Thank you received: 559
Although I like the plate solve method, I do not want to abandon the move star method, because as is obvious, plate solving can be problematic. Move star it's probably better for slower CPUs.

So, for move star, fast refresh times and different exposure times between the first 3 captures and the refresh captures are important. I don't want to constrain both exposures to be identical. Perhaps the field should be labeled refresh exposure?

In neither case do I see a reason to delay. If you happen to move the scope during a capture and get a bad image, why not just ignore that result?
1 year 6 months ago #86144

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

  • Posts: 1309
  • Thank you received: 226
I see no need to change the refresh function for the legacy move star, or move star + calc. I'm only advocating this for the plate solve method.
My reason for that is I have not experienced much in the way of a graceful recovery from bad frames. I've found bad frames either cause significant time-out delay with little information as to what is taking place while it is trying to process it. Or worse case scenario, it gets hung up in the process and never takes a new clean frame, requiring me to abort and redo the entire process, if I'm lucky it will let me. If I'm not lucky I have to restart the drivers.
1 year 6 months ago #86147

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

  • Posts: 17
  • Thank you received: 4
The timing is not the problem....

The plate solve refresh has a huge advantage for small sensors with large focal length, leading to very narrow FOV. I only have less than 0,5° FOV and have to setup and align my EQ6 every time i want to work with it. Normally my initial polar alignment ist 1-2° away. In this case the "Move Star" method is nearly useless as the triangle does not fit in the sensor window. And: I need to make large correction to the alignment screws and the "move star" refresh often looses the star as it moves out of the window.

Plate solve PA would be a great help for users like me - regardless of the refresh cycle duration.

I normally do PA near Polaris. And the plate solving of the first 3 images (and slew to target, sync) work perfectly even for my small FOV and even in regions with only a few stars like around Polaris. ONLY the refresh does not work and the main problem is that there is no meaningful error message.

To reproduce it you can use the following simulator settings:

Focal length: 1000 mm
Sensor dimensions: 11,3 mm x 11,3 mm
Resolution: 3008x3008 px
Binning 2x2 (to reduce the amount of data and the size of the captured image)
Pointing the scope to polaris and start Polar Align with 20° or 30° delta to east or west.

This should be enough info to reproduce the problem.

Exposure time varies from 2 to 10 s - the result is always the same: Normal platesolving works. First 3 images during PA solve perfectly.
Refreah phase: Plate solve failed, no entry in the solver list, no meaningful error message.

Tested on a RPi4 with 2 GB RAM and on a windows PC coupled to INDI running on the RPI4.

Ciao, Udo
Last edit: 1 year 6 months ago by Udo.
1 year 6 months ago #86148

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

  • Posts: 333
  • Thank you received: 23
Udo, you reproduced perfectly the situation

1 year 6 months ago #86149

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

  • Posts: 1208
  • Thank you received: 559
Udo,

I ran the simulator using your specs and had no issues. See this movie.
I tried it pointed right at Polaris, as in the movie attached, and 5 degrees east of Polaris. In both cases it succeeded.
Did the simulator fail for you? Note: you do not enter sensor dimension in the simulator/INDI control panel, so I simply divided 11.3 mm / 3008 px to get a 3.75u pixel size.
Please try and replicate it on your simulator, and send me the specifics of how you set it up.

Hy
1 year 6 months ago #86151
Attachments:

    simplescre...2.36.mkv

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

  • Posts: 17
  • Thank you received: 4
I will try again tomorrow and add screenshots of my solver settings.
What hardware/CPU did you use?

Ciao, Udo
1 year 6 months ago #86152

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

  • Posts: 1208
  • Thank you received: 559
I used my mac laptop running an Ubuntu VM.
It's best for me if you can re-create this issue with the simulator, if possible.
Can you compile and run code? If so, I can show you how to, e.g. save the refresh images as fits files.
Hy
1 year 6 months ago #86155

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

Time to create page: 1.400 seconds