Pixinsight registration failed

[Leti Theobald]

Hello all,

Having lots of issues stacking in Pixinsight and can't figure out why. I've looked at my images and they are fine but still failing to register and rejecting most of my images.

I've attached the log, any ideas anyone? Could I have ticked something I should not? I've restored to factory defaults just in case but same result

ProcessLogger.txt

Edited January 19 by Leti Theobald

[pipnina]

Hi

It's hard to diagnose this issue without more info.

If you can supply 3+ of your raw images (of the same colour) I can load them into my PixInsight and see if the images need tweaked settings to register properly.

As a shot in the dark: Have you tried registering them manually as a test of the registering setting by going to "process-> image registration -> star alignment" ?

Load only 2/3 images and test it on those. change one setting at a time and once you find a good setting use that in the WBPP script.

[Leti Theobald]

3 minutes ago, pipnina said:

Hi

It's hard to diagnose this issue without more info.

If you can supply 3+ of your raw images (of the same colour) I can load them into my PixInsight and see if the images need tweaked settings to register properly.

As a shot in the dark: Have you tried registering them manually as a test of the registering setting by going to "process-> image registration -> star alignment" ?

Load only 2/3 images and test it on those. change one setting at a time and once you find a good setting use that in the WBPP script.

Ok, had a go and this is the message I get. I've attached 3 images for you

Thank you so much, I've never had issues stacking with Pixinsight before

 

Light_M101_300.0s_Bin1_B_gain0_20240118-045759_-10.1C_0030.fit Light_M101_300.0s_Bin1_B_gain0_20240118-045257_-10.5C_0029.fit Light_M101_300.0s_Bin1_B_gain0_20240118-044725_-10.0C_0028.fit

[Leti Theobald]

Also, this are the parameters I got

[pipnina]

I believe your issue is that your images simply aren't close to focus! PixInsight likely cannot recognise the stars in your image as stars as a result.

For instance this bright star in an unstretched image near the center of the frame. This donut shape is typical of the pattern given when a reflector telescope is out of focus (the point of light widens into a circle with a harsh border in lenses, called bokeh, however due to the central obstruction in reflectors a hole appears in the middle creating these frustrating donuts!)

I used the Resample process in Pix to reduce the image size to 25%, thereby making the stars more star-shaped. They then registered immediately with the settings you see in the screenshot.

I hope this helps!

[Leti Theobald]

Thank you, I'll try that, though not great that is out of focus! I have an autofocuser so weird that is out of focus, maybe I need it to refocus more often through the night

[pipnina]

Just now, Leti Theobald said:

Thank you, I'll try that, though not great that is out of focus! I have an autofocuser so weird that is out of focus, maybe I need it to refocus more often through the night

Perhaps, but I have found autofocus routines need to be relatively close to good focus in order to be effective, especially when I owned a newtonian reflector.

I would suggest manually getting focus close, and then running the autofocus routine and seeing how it handles the scene, does its output look correct in a quick test image? etc.

Try and avoid the step-out that it might use taking you too far from ideal focus as a starting point.

Unlike mirrorless cameras which can use phase-detect focusing and at long focal lengths could provide fast and accurate focusing information, we are purely limited to contrast-based AF which is very slow and prone to errors. We need to experiment with this as astrophotographers and find out where the limits are.

For instance in my setup, if ideal focus was at step count 8385, and the current focus position was at 8320, then running the autofocus routine might move focus out to 8820 and start working back down in 100 step intervals. This could easily prove too much error from the ideal position by creating large donut stars that the computer can't identify properly, leading to a bogus AF result.

However if you tune the step-out multiple to be lower, and lower the step interval to maybe 50 steps, it might move from 8385 to 8500 and be able to focus properly.

It's worth bearing in mind that the only movement in your imaging train when it comes to focus postion comes from thermal expansion and switching filters, so the real difference between the highest and lowest authentic result is not big- could be as much as 100 steps between a very cold night (in the southern UK I define that at -5c) and a very warm night (in UK i'd say 20c).

It's all about tuning that eventually allows you to just let the AF do its thing with confidence that it has worked. Personally though I always stick around for the first image to come back at least before letting it go by unattended!

I wish you the best of luck with the images you currently have, and the no doubt far improved result you will get next time!

[Leti Theobald]

8 minutes ago, pipnina said:

Perhaps, but I have found autofocus routines need to be relatively close to good focus in order to be effective, especially when I owned a newtonian reflector.

I would suggest manually getting focus close, and then running the autofocus routine and seeing how it handles the scene, does its output look correct in a quick test image? etc.

Try and avoid the step-out that it might use taking you too far from ideal focus as a starting point.

Unlike mirrorless cameras which can use phase-detect focusing and at long focal lengths could provide fast and accurate focusing information, we are purely limited to contrast-based AF which is very slow and prone to errors. We need to experiment with this as astrophotographers and find out where the limits are.

For instance in my setup, if ideal focus was at step count 8385, and the current focus position was at 8320, then running the autofocus routine might move focus out to 8820 and start working back down in 100 step intervals. This could easily prove too much error from the ideal position by creating large donut stars that the computer can't identify properly, leading to a bogus AF result.

However if you tune the step-out multiple to be lower, and lower the step interval to maybe 50 steps, it might move from 8385 to 8500 and be able to focus properly.

It's worth bearing in mind that the only movement in your imaging train when it comes to focus postion comes from thermal expansion and switching filters, so the real difference between the highest and lowest authentic result is not big- could be as much as 100 steps between a very cold night (in the southern UK I define that at -5c) and a very warm night (in UK i'd say 20c).

It's all about tuning that eventually allows you to just let the AF do its thing with confidence that it has worked. Personally though I always stick around for the first image to come back at least before letting it go by unattended!

I wish you the best of luck with the images you currently have, and the no doubt far improved result you will get next time!

Thank you so much for the tips, I will have a look at my auto focuser settings. I'm not that experience so super useful advice

[tomato]

If your imaging software records stats like the star count and mean FWHM for each image, you can look at these to see how the quality of the subs is going. Generally my targets start lower in the sky and move up towards the meridian, so the star count increases and usually the FWHM value decreases, but conditions can alter. In NINA I request a refocus if the FWHM degrades by more than 10% of the initial value.