Great video. Thanks for all your detailed videos and incredible tools - this is an amazing channel and you're singlehandedly making such a big difference
This was really helpful, Frank. Thank you. I bit the bullet tonight & installed 1.9.2, not without trepidation because I found installing one of the 1.8 versions traumatic. But it worked on my Windows 10 PC and, thanks to you, I found everything.
Holy cow! 🤯 Thank you so much for this. I have been battling since 1.9 launch to get it to run stable on my computer, every time I ran WBPP or FBPP it simply locked and couldn't do anything else, the thread priority setting change seems like the smoking gun for that behavior.
Thanks Frank, a couple of good hints that were new for me. The 1.9 feature that intrigued me most was SPFC and the new MultiScaleGradient correction tool with the MARS data. I got caught out by forgetting that the MARS database is currently incomplete (specifically very little Oiii support). Initially I couldn't figure out why the 'hole' in the Rosette was darker than the corrected background ... and then the penny dropped. (thanks to community reminders LOL). Love your scripts btw! PS just found your other vid on MGC 👍
I'm new to PI and 1.9 is all I've know. That said, it has crashed at least 3 times in last two weeks. I just changed the setting from time critical to highest and will see if that helps.
Great information. Thank you! I found the speed not to be an issue for me. However, the v1.9 takes longer on my new M4 Mac Mini with 32GB of RAM. I am hoping this will work itself out over the next few weeks. Also, I was seeing the bouncing beach ball for the first time. I changed the settings you suggested, and hopefully, I will get back to normal.
Cool tip regarding the 'General Extreme Studentised Deviate' integration method. Just had to change my Interpolation method under Star Alignment to get better results, which led me to realise i know next to nothing on why or when to use certain algorithms. Could make a cool video if you know tips in these steps, Calibration, Star Alignment, Integration, and even Drizzle Integration.
Thank you! I like using my computer, including video and voice comms, while PI works in the background. I set the max thread priority to lowest then I can just let it rip and just check in once in a while to see how it's going. This, for me, works better than limiting the number of threads. YMMV
Thanks! I bought Pixinsight because it makes high quality images. Some people report PI out performs the other out there. I can’t understand why people want things faster with less quality. We endure a lot to acquire data. 😅Why would we want to compromise in processing just for speed?
I think one aspect is that people are integrating more and more frames and at a certain point the rejection methods no longer make a difference. The law of large numbers comes into play and it makes sense to favor speed because it is no longer at the expense of quality.
@ I didn’t think of that. I’m using more and more frames in a stack. Thanks for this information. When you think of it that way, maybe it doesn’t make sense. Still, I don’t think it should be default ON fast mode.
Just on 3:30, the default Rejection Algorithm for mine 1.9.2 /WBPP v2.8.2 seems to be Auto (which states that'll it'll then select the appropriate algorithm based on number of subs). However I don't know if it's documented what the cutoff number of subs is for each rejection algorithm.
Great tutorial as always! This and all of the tools you've been creating have been very helpful in my processing. One question. I recently updated to 1.9 and my WBPP is still on v2.8.0 and I noticed you are on v2.8.2. In my version 2.8 I cannot make adjustments to the Calibration menus along the right side. Am I missing something or can we not adjust them? I am new to using WBPP so hence the confusion. Thanks
What do you do if it won't install? There is no support since I have reached out to the forum and the authors and gotten no response whatsoever from either group. I have had it installed on both computers using past versions with no issues. It flashes text for a split second and then only a gray rectangle with the blue cube.
on my asi1600mm i drizzled every image every time. 2x scaling. on my new asi294mm the bin 1 is the arcsec/pixel i am shooting for so havent drizzled that data yet
Thanks Frank, appreciate you covering the update quirks and optimizations! I didn’t buy PI because it was fast 😂. I generally could care less about its speed and am more concerned with quality and stability. Cheers! Doug
Not sure which one of the settings I applied but it’s punching holes in my stars. I suspect it’s from changing drizzle settings from Square to Gaussian. Will have to play around with it to confirm
Your cpu needs to have AVX2 instruction set. Just google if your does and there are some ways you can check. You can use somethibg like cpuid's cpu-z freeware to look too
Every time I update PI it breaks the connection to StarNet2. I usually just reinstall the module, but now I get a message that says… failed. Error: required module signature, not found: no modules were installed. Is this also something new for version 1.9.2? I’ve always been able to reinstall Starnet2 before. If you ask me,StarNet2 should be part of PI to begin with. Anyone able to get the starnet2 module installed?
Btw. with Redcat 51 and 2600MC Pro, would you recommend Drizzle shrink at 0.9? I was always doing default 1.0 When you hover over that setting it says default is 0.9, but if you reset to default settings it's at 1.0... lol
Yes. Juan, the head developer for PixInsight wrote this: "The software development world is complicated, especially when working on a large and complex platform like PixInsight. Sometimes, we must make delicate decisions that are not always guided by strictly technical reasons, and sometimes, we make mistakes that have a cost in a variety of ways. GESD is an excellent outlier detection algorithm, as verified in the standard literature on statistics and in practice when you use it as a pixel rejection algorithm with our ImageIntegration process. However, despite having an optimal implementation, GESD requires significant CPU resources, which makes its use problematic for us in practical marketing terms. To summarize, the decision to use GESD by default in WBPP was a mistake that caused us much harm during the past years. Most people in the astrophotography world think that PixInsight's preprocessing is overly slow and resource-consuming, which has given a significant advantage to our worst competitors. Actually, it is just the opposite: our FastIntegration process and the FBPP script (or equivalently, WBPP with the appropriate selection of parameters) have unparalleled fast execution and efficiency. But who knows that these processes and scripts exist or how they can be configured for optimal performance? Few people. Unfortunately, most people in the amateur astrophotography world only want to do one click to get instant results, with basically zero interest in anything related to image processing, let alone knowledge about algorithms, rigorous data reduction and processing methodologies, and philosophical considerations. All of that has little value today. Of course, PixInsight will never be like that and will always aim to achieve excellence, but if we must survive, we must inevitably make some concessions. One of them is changing the default rejection algorithm in WBPP to linear fit clipping instead of GESD. LFC is a very good rejection algorithm, works well with moderate to large data sets, and is fast. It is helping us remove our bad reputation of being slow, which is necessary to fight against our worst freeware competitors, constantly trying to copy and replace us shamelessly."
I seem to remember that Adam Block mentioned the same and he too could not understand why they changed that default value at PI. Maybe he will come back on this one in a later video in his series. Anyway, I'll wait some more time before installing 1.9.x as there may be some other bugfixes waiting...
Yes, that was a note to all use 3rd party developers as to the reason for the change.... they should've had something about that in the release notes that the default was changed to ease the burden on computation time and GESD is still an option for optimal rejection or something to that affect
Great video.
Thanks for all your detailed videos and incredible tools - this is an amazing channel and you're singlehandedly making such a big difference
Thank you so much!!!
This was really helpful, Frank. Thank you. I bit the bullet tonight & installed 1.9.2, not without trepidation because I found installing one of the 1.8 versions traumatic. But it worked on my Windows 10 PC and, thanks to you, I found everything.
This is extremely helpful, thanks for bringing it up!
I'm learning every day... Thanks again.
Great review as always - many thanks!!!
Holy cow! 🤯 Thank you so much for this. I have been battling since 1.9 launch to get it to run stable on my computer, every time I ran WBPP or FBPP it simply locked and couldn't do anything else, the thread priority setting change seems like the smoking gun for that behavior.
awesome to hear!!
Great video. Thank you, greatly appreciated.
Thanks Frank, a couple of good hints that were new for me. The 1.9 feature that intrigued me most was SPFC and the new MultiScaleGradient correction tool with the MARS data. I got caught out by forgetting that the MARS database is currently incomplete (specifically very little Oiii support). Initially I couldn't figure out why the 'hole' in the Rosette was darker than the corrected background ... and then the penny dropped. (thanks to community reminders LOL). Love your scripts btw! PS just found your other vid on MGC 👍
Thank you for the detailed insight into this new release, 1.9. Very Helpful.
Thanks!
Incredibly helpful. Thanks for this!
I'm new to PI and 1.9 is all I've know. That said, it has crashed at least 3 times in last two weeks. I just changed the setting from time critical to highest and will see if that helps.
Be sure to grab my script repository, it will help speed up your processing.
Thank you
Great information. Thank you! I found the speed not to be an issue for me. However, the v1.9 takes longer on my new M4 Mac Mini with 32GB of RAM. I am hoping this will work itself out over the next few weeks. Also, I was seeing the bouncing beach ball for the first time. I changed the settings you suggested, and hopefully, I will get back to normal.
Thx Frank
Helpful Frank!
Cool tip regarding the 'General Extreme Studentised Deviate' integration method. Just had to change my Interpolation method under Star Alignment to get better results, which led me to realise i know next to nothing on why or when to use certain algorithms. Could make a cool video if you know tips in these steps, Calibration, Star Alignment, Integration, and even Drizzle Integration.
I think the view synchronisation is one of the best new features. Also the copy astrometric solution between views is now also just drag and drop.
Yes! And copy stf from one to another. My last pixinsight video covered those items too
Thank you!
I like using my computer, including video and voice comms, while PI works in the background. I set the max thread priority to lowest then I can just let it rip and just check in once in a while to see how it's going. This, for me, works better than limiting the number of threads. YMMV
Great video! Very helpful!
Thanks! I bought Pixinsight because it makes high quality images. Some people report PI out performs the other out there. I can’t understand why people want things faster with less quality. We endure a lot to acquire data. 😅Why would we want to compromise in processing just for speed?
I think one aspect is that people are integrating more and more frames and at a certain point the rejection methods no longer make a difference. The law of large numbers comes into play and it makes sense to favor speed because it is no longer at the expense of quality.
@ I didn’t think of that. I’m using more and more frames in a stack. Thanks for this information. When you think of it that way, maybe it doesn’t make sense. Still, I don’t think it should be default ON fast mode.
Just on 3:30, the default Rejection Algorithm for mine 1.9.2 /WBPP v2.8.2 seems to be Auto (which states that'll it'll then select the appropriate algorithm based on number of subs). However I don't know if it's documented what the cutoff number of subs is for each rejection algorithm.
In the WBPP section of the announcement, Juan mentioned that the default rejection method for >15 frames has been changed to LinearFitClipping.
Great tutorial as always! This and all of the tools you've been creating have been very helpful in my processing. One question. I recently updated to 1.9 and my WBPP is still on v2.8.0 and I noticed you are on v2.8.2. In my version 2.8 I cannot make adjustments to the Calibration menus along the right side. Am I missing something or can we not adjust them? I am new to using WBPP so hence the confusion. Thanks
Are you on v1.9.2? That is the newest PI out right now
What do you do if it won't install? There is no support since I have reached out to the forum and the authors and gotten no response whatsoever from either group. I have had it installed on both computers using past versions with no issues. It flashes text for a split second and then only a gray rectangle with the blue cube.
does your computer meet the new hardware requirements?
Another great, very informative video, Franklin. Thanks again.
BTW, do you Drizzle? And if so, do you ever change the Scale Factor?
on my asi1600mm i drizzled every image every time. 2x scaling. on my new asi294mm the bin 1 is the arcsec/pixel i am shooting for so havent drizzled that data yet
Thanks Frank, appreciate you covering the update quirks and optimizations! I didn’t buy PI because it was fast 😂. I generally could care less about its speed and am more concerned with quality and stability.
Cheers!
Doug
Agreed!
Thanks for the heads up! IMO, PixInsight has become unnecessarily complicated.
And i was wondering, why my last stack (400+ subs) was so fast. Don't like the changes. I choose quality over speed anytime. Thanks for the video!
Not sure which one of the settings I applied but it’s punching holes in my stars. I suspect it’s from changing drizzle settings from Square to Gaussian. Will have to play around with it to confirm
No, it is the automatic Cosmetic Correction. Turn that back to manual and either ise your saves CC template or leave it off!
I've tried to download the latest version on my laptop but I can't get it to work. I assume it's because my laptop is too old for the newest version??
Your cpu needs to have AVX2 instruction set. Just google if your does and there are some ways you can check. You can use somethibg like cpuid's cpu-z freeware to look too
If I drag the triangle from WBPP to create an icon for it, then save that icon, is the cache saved with that icon?
No. If you clear your cache it is gone
Every time I update PI it breaks the connection to StarNet2. I usually just reinstall the module, but now I get a message that says… failed. Error: required module signature, not found: no modules were installed.
Is this also something new for version 1.9.2? I’ve always been able to reinstall Starnet2 before. If you ask me,StarNet2 should be part of PI to begin with. Anyone able to get the starnet2 module installed?
You have to download the new version of starnet for this v1.9 of PI
Good video, thanks.
I hate that they try to make everything so "fast" by sacrificing quality of the output image.
Btw. with Redcat 51 and 2600MC Pro, would you recommend Drizzle shrink at 0.9? I was always doing default 1.0
When you hover over that setting it says default is 0.9, but if you reset to default settings it's at 1.0... lol
2:58 any sources confirming this?
Yes. Juan, the head developer for PixInsight wrote this:
"The software development world is complicated, especially when working on a large and complex platform like PixInsight. Sometimes, we must make delicate decisions that are not always guided by strictly technical reasons, and sometimes, we make mistakes that have a cost in a variety of ways.
GESD is an excellent outlier detection algorithm, as verified in the standard literature on statistics and in practice when you use it as a pixel rejection algorithm with our ImageIntegration process. However, despite having an optimal implementation, GESD requires significant CPU resources, which makes its use problematic for us in practical marketing terms. To summarize, the decision to use GESD by default in WBPP was a mistake that caused us much harm during the past years. Most people in the astrophotography world think that PixInsight's preprocessing is overly slow and resource-consuming, which has given a significant advantage to our worst competitors. Actually, it is just the opposite: our FastIntegration process and the FBPP script (or equivalently, WBPP with the appropriate selection of parameters) have unparalleled fast execution and efficiency. But who knows that these processes and scripts exist or how they can be configured for optimal performance? Few people.
Unfortunately, most people in the amateur astrophotography world only want to do one click to get instant results, with basically zero interest in anything related to image processing, let alone knowledge about algorithms, rigorous data reduction and processing methodologies, and philosophical considerations. All of that has little value today. Of course, PixInsight will never be like that and will always aim to achieve excellence, but if we must survive, we must inevitably make some concessions. One of them is changing the default rejection algorithm in WBPP to linear fit clipping instead of GESD. LFC is a very good rejection algorithm, works well with moderate to large data sets, and is fast. It is helping us remove our bad reputation of being slow, which is necessary to fight against our worst freeware competitors, constantly trying to copy and replace us shamelessly."
I seem to remember that Adam Block mentioned the same and he too could not understand why they changed that default value at PI. Maybe he will come back on this one in a later video in his series. Anyway, I'll wait some more time before installing 1.9.x as there may be some other bugfixes waiting...
Yes, that was a note to all use 3rd party developers as to the reason for the change.... they should've had something about that in the release notes that the default was changed to ease the burden on computation time and GESD is still an option for optimal rejection or something to that affect
Funny… if they want better performance, allow more multicore processing then reducing the function pls.
Thanks!
Very useful video, thanks a lot!