Process Console data sequence incorrect during Star Alighnment

[Tony Morris]

Hi
I have been using the star alignment data in the process console to characterise the periodic error in the worm and wheel assembly in my mount.
However the data generated in the console window is not in the same sequance as the image order as correctly displayed in the star alignment process window.
My images are ordered by file name.
I have tried prefixing my files with 01,02 etc but this does not help.
If I try a different set of files the same problem occurs, but different files are processed out of sequence.
I am running PI 01.08.05.1352 on Xubuntu 18.04
Here is a console output that shows files not being loaded in sequence:-

StarAlignment
Loading reference image
IMG_3502.CR2
Registration of 11 target files.
* Using 10 worker threads.
[000] IMG_3502.CR2
[001] IMG_3503.CR2
[002] IMG_3504.CR2
[003] IMG_3505.CR2
[004] IMG_3506.CR2
[005] IMG_3507.CR2
[006] IMG_3508.CR2
[007] IMG_3509.CR2
[008] IMG_3510.CR2
[009] IMG_3511.CR2
Loading target file
IMG_3502.CR2
Loading target file
IMG_3503.CR2
Loading target file
IMG_3504.CR2
Loading target file
IMG_3506.CR2
Loading target file
IMG_3505.CR2
Loading target file
IMG_3507.CR2
Loading target file
IMG_3509.CR2
Loading target file
IMG_3508.CR2
Loading target file
IMG_3512.CR2
Loading target file
IMG_3510.CR2
Loading target file
IMG_3511.CR2
Registration successful.

Is there a setting I am missing to force the order or is this a bug?
Thanks
Tony

[msmythers]

I would think there is no requirement of the tool to process the images in any order. The reference image is the only set requirement I think. Now with that said you could try turning off parallel processing in the Global Preferences. This will greatly increase the time to process the images but it might change the order of processing.



Mike

[RickS]

Mike is correct. PI is using 10 different threads, each registering a separate image in parallel.  The order they finish doesn't matter. You can turn off parallel processing but that will make everything run slower.

[Tony Morris]

Ok I can see the argument that the star alignment tool processing sequence does not matter (for star alignment) but there is a lot of very useful data in the process console and would be more user friendly if it was in the correct file or image order.
It's certainly a pain cutting and pasting the date back into the correct sequence for analysis
PI is very logical & robust in all that it does, this quirk seems a little odd (maybe random) to a non programmer like myself.

Tony

[RickS]

PI is very logical & robust in all that it does, this quirk seems a little odd (maybe random) to a non programmer like myself.

And it seems perfectly natural to someone like me who was once a software engineer before I grew the pointy hair

It is purely about wringing the most performance out of multiple CPU cores/threads.  You want them all working simultaneously as much as possible.  It would have been possible to present the results in the original order but only with extra synchronization/programming work.

What sort of analysis do you do that requires cutting & pasting the log back together in order?  I skim the output to look for issues and inspect it more deeply if there's a problem, but I've never felt the need to pore through it like that.

Cheers,
Rick.

[Tony Morris]

Hi Rick

I don't want to loose the benefit of multiple threads, not on a old machine like mine!!!

I have started to use the dx and dy pixel displacement data that is generated during star alignment between frames on my unguided imaging runs.
I copy and paste the console information into a spreadsheet and graph the dx and dy data points.
The console data has to be sequenced correctly, i/e in the same sequence as the frames are taken. So far I have generated a rough sinusoid with noise, if any of the data points are out of sequence its relatively easy to spot in the graph.
I think my graph is telling me my worm and wheel  needs sorting but that's another story.
If PI is developed with more mount and camera control this type of data is useful for tuning before issues are hidden by auto-guiding

Tony

[RickS]

I have started to use the dx and dy pixel displacement data that is generated during star alignment between frames on my unguided imaging runs.
I copy and paste the console information into a spreadsheet and graph the dx and dy data points.
The console data has to be sequenced correctly, i/e in the same sequence as the frames are taken. So far I have generated a rough sinusoid with noise, if any of the data points are out of sequence its relatively easy to spot in the graph.
I think my graph is telling me my worm and wheel  needs sorting but that's another story.
If PI is developed with more mount and camera control this type of data is useful for tuning before issues are hidden by auto-guiding

Thanks for the explanation, Tony.  That does sound like a good use for sorted output.

It looks like your image files are named in sequence so perhaps you can sort on the file name in the spreadsheet and save some manual effort that way?  Shame you're not a programmer or you could write a little script to extract and sort the data.

Cheers,
Rick.