Bug has been confirmed by Juan on windows and OSX. I have several friends who also have this issue (so this is not my computer(s)
Juan, if you can answer ?
PixInsight 1.8.8-10 Released
- Thread starter Juan Conejero
- Start date
Bug has been confirmed by Juan on windows and OSX. I have several friends who also have this issue (so this is not my computer(s)
Juan, if you can answer ?
Hi Jaun,
The SS problem I mentioned above was either a glitch or has been fixed with the recent update (SS v1.6.1). Thanks either way.
I am sure this is a dumb question, but I'm a bit confused as to which indicates the best frame; a high or a low "PFS Signal Weight"?
So in the image is image Index 6 the best or worst frame? It certainly has the lowest star count, so I assume its the worst.
Thanks, Jim
The SS problem I mentioned above was either a glitch or has been fixed with the recent update (SS v1.6.1). Thanks either way.
I am sure this is a dumb question, but I'm a bit confused as to which indicates the best frame; a high or a low "PFS Signal Weight"?
So in the image is image Index 6 the best or worst frame? It certainly has the lowest star count, so I assume its the worst.
Thanks, Jim
The reported issues with the SubframeSelector process have been fixed with an update to version 1.6.1 of the SubframeSelector module.
As for image weights, the higher the better.
As for image weights, the higher the better.
JPetruzzi
Well-known member
Something a little curious--I had excellent success w/the new subframe selector playing with several data sets yesterday...however after doing the latest update this morning I ran another analysis and the frame that got the highest PFS signal weight was this frame that had a serious tracking error, which surprised me. Is this an outlier frame because of near-doubled stars? Just thought I'd run it past you.
The PSF weighting algorithm has not been designed to detect tracking errors or double exposures, as happens with this frame. Actually, the algorithm is quite robust to these errors. It will estimate the mean signal of the image from all detected stars, by fitting point spread functions and measuring the intensities of the pixels above the local background for each of them. The dimmest 'companion' of each 'double' star will be ignored by the PSF fitting process, although the double star images will probably contaminate each sample to a relatively small extent.
In other words, In general, PSF signal estimation cannot be used to detect these anomalies. To filter out these bad frames you should use other metrics, such as eccentricity for example, or visual inspection with the Blink tool.
In other words, In general, PSF signal estimation cannot be used to detect these anomalies. To filter out these bad frames you should use other metrics, such as eccentricity for example, or visual inspection with the Blink tool.
JPetruzzi
Well-known member
Yes, I always blink my images as well--I think this is the only situation I've seen after re-running 5 old datasets that was not a perfect match between "best frame" and PSFSW. Overall I remain REALLY impressed with the new PSF implementation. Thanks for taking the time!
Isbeorn
Member
A nice set of additions!
Especially the NetworkService Module seems to be something I could utilize as soon as the docs are available for it.
The addition to embed the processing history into the XISF file is such a good idea. The XISF format is already a great format and adding the history to a finalized image to remember how an image was processed is perfect.
Thank you.
Especially the NetworkService Module seems to be something I could utilize as soon as the docs are available for it.
The addition to embed the processing history into the XISF file is such a good idea. The XISF format is already a great format and adding the history to a finalized image to remember how an image was processed is perfect.
Thank you.
jmurphy
Well-known member
I agree, this new feature is fantastic. Thank you @Juan Conejero !
I have already made PhotometricMosaic compatible with this, and I am currently adding the ability to NSG.
I have just published an introductory document where I describe everything you need to start working with NetworkService:
The NetworkService process in PixInsight 1.8.8-10
The NetworkService module implements the necessary software infrastructure to run PixInsight in a fully automated environment as an on-demand service. This module is at the heart of our AstroBin+PixInsight collaboration, where PixInsight is constantly providing accurate and feature-rich...
pixinsight.com
Let me know if you need more information.
Christian_Gloor
New member
Hello everyone,
I must be missing something... I'm running the latest version of Pixinsight and WBPP on macOS 10.15.7.
I wanted to test out the new weighting method on my data but when using the default "PSF signal" in the "Subframe Weighting" section, the WBPPWGHT value is not written in the header of the xisf file.
If I chose "Weighting formula", it works as it used in the past and WBBWGHT is set, allowing me to use it when manually stacking, as it used to work in the past.
Is this behavior expected?
I must be missing something... I'm running the latest version of Pixinsight and WBPP on macOS 10.15.7.
I wanted to test out the new weighting method on my data but when using the default "PSF signal" in the "Subframe Weighting" section, the WBPPWGHT value is not written in the header of the xisf file.
If I chose "Weighting formula", it works as it used in the past and WBBWGHT is set, allowing me to use it when manually stacking, as it used to work in the past.
Is this behavior expected?
Yes, this is the expected behavior. When you select one of the new weighting methods (PSF Signal, PSF Power or SNR Estimate), the WBPP script does not generate weight keywords because they are not necessary, since all the required metadata is already included in the calibrated and registered frames. In other words, the ImageIntegration tool already knows how to calculate the weights if the data have been calibrated with the new version of PixInsight. Just leave ImageIntegration with its default PSF Signal weighting option (or select PSF Power / SNR estimate).
Just a word of caution: The SNR estimate method is sensitive to outliers, so always use it after verification with the SubframeSelector tool (read the announcement post for an introductory description).
Just a word of caution: The SNR estimate method is sensitive to outliers, so always use it after verification with the SubframeSelector tool (read the announcement post for an introductory description).
Christian_Gloor
New member
Excellent!.
Thank you Juan for the quick answer. Now I understand and it makes sense.
Great work!
crackout
Well-known member
Yes, correct. I meant the notation on the graph.
crackout
Well-known member
I have another question:
I've been using external PSFs for Deconvolution until now with good to great success.
My pixel scale is 1.88''/px and the PSFs came out pretty blocky. However, they worked fine with GlobalDark around 0.003.
With the new version, my PSFs look a lot more oversampled:
This results in heavy ringing using my usual settings.
I haven't been able to replicate the nice results of earlier PI versions until now.
I've been using external PSFs for Deconvolution until now with good to great success.
My pixel scale is 1.88''/px and the PSFs came out pretty blocky. However, they worked fine with GlobalDark around 0.003.
With the new version, my PSFs look a lot more oversampled:
This results in heavy ringing using my usual settings.
I haven't been able to replicate the nice results of earlier PI versions until now.
Nothing has changed in the Deconvolution tool for a long time, so the cause of the changes you are observing must be elsewhere. On the other hand, the performance of a PSF image does not depend on how you see it represented on the screen. It's its shape and dimensions what matters.