Thanks Shawn. In is documentation , Russel write : "if Automatic PSF is selected, any aberration correction (coma, astigmatism, etc.) applied to stars will also be applied to nonstellar features. This is not true in manual mode (unless the Correct First option is selected):the PSF of nonstellar features will be assumed to be round." So maybe, it will be beter to select "correct First" option when you use manual mode? Sorry for my english, i'm french..😉
Since I had completely avoided convolution completely, there was no struggle. ;-) BXT impact is something that may or may not be that noticeable at "screen size" viewing. But its impact becomes more apparent as the image size increases. In my M45 that was posted earlier this week, I applied BXT as I was just about through with the image. When zoomed in the increased detail around the primary stars and their shock waves in the dust cloud have noticeable improvements in detail/contrast.
Great work Shawn! - The tool itself has invigorated me for astronomy even more than I already was, a true game changer I reckon! The PSF script looks really handy by the way, thanks for the tip!! 🙂 Clear skies buddy, hope you and yours have a lovely christmas and new year!
I bought it the minute I saw it a few days ago. I have been having a reprocessing marathon and the results are nothing short of amazing. Thanks for this video as everything you guys post is appreciated.
Fantastic, what a result. Thanks Shawn, another great presentation. Having struggled with deconvolution for too long this is going to be my next purchase.
Great stuff! Bought it as soon as I saw it. It's funny that I was just recently wondering if Russ could do something with Deconvolution! I had mostly given up despite your great tutorial. Thanks also for the tip on PSFimage too. I did not have that one yet. The few images I tried so far I used this after SPCC but will move it up to post DBE now. Cheers!
It’s a no-brainier and a game-changer… I’m wanting to frame my images zoomed in more now I have all this extra detail. I’ve done a few re-edits and the one thing I have discovered is that it’s easy to over-process now… BlurX does such an awesome job that I think my workflow will be a lot less after combining than it was prior… it’s amazing 👍🏻
Sean, Just started my processing after WBPP of M42 (LSHO). I set up the Ha image with two previews. On the first I ran BXT at defaults, then I ran BXT with the PSF determined by PSF analyzer. I as blown away by the difference, especially in cloud details. It is definitely worth running the PSF analysis and taking that chore off of the BXT. I have to say that the RUclips videos really don't do justice to the impact of BXT, especially with the PSF 'tuned' in.
Thanks for the video. I am just now trying the BlurXterminator on the heart nebula with fantastic results. One thing that I notice is that the process is quite slower than what I saw in your video. I have a really poweful computer game, top cpu, memory (32GB), etc; and still we take a good time to complete.
I have the Nvidia hack done so that blurX and noiseX use the GPU to perform their tasks. Not the CPU which is slower. It's a bit involved to do the hack but it does work well.
@@alexalmeida8627 Did you follow William Li's method? There was also one additional file that needed to be installed (according to Russ) to make it work. This was for StarX but may also help with speeding up BXT. If you get in touch with me via email (see my website visibledark.ca for contact), I can send you the links.
Gosh Darn it Shawn! I was trying to avoid watching this video but finally broke down, your results speak for for themselves. I have Russ's other products and figured I would probably have to have this one as well. Thanks for another excellent PixInsight tutorial! Cheers Kurt
Was reading the document for Blurxterminator and noticed a comment about diffraction spikes that I get with my reflector. Not sure if it was saying it would correct for those? In some pictures, I like the dramatic affect of the spikes. Definitely going to purchase this. And loved all the comments about workflow.
I got trial version few days ago.Tool is absolutely amazing.No more straggling with deconvolution 🙂.Thanks for sharing tips how to further improve image
So funny. I went to Cloudy Nights this weekend asking for some help reducing star size with open star clusters and someone recommended this. I took one look at it and just bought it. Like fantastic.
I've never used deconvolution, maybe now I'll do it with this program. It always seem to make a little difference for a lot of trouble when I tried it.
Give it a try. Great results with BXT. And makes decon a lot easier. I actually wouldn't do decon (the old way) for some of my images because it was to much work! lol. Now BXT changes all that.
Hey Julian - sorry for the delay. I get swamped with questions and it is difficult to answer all of them and timely. I press control-shift and Z to do it. It can only be done on previews.
So far you have the best video on BXT . Couple of questions . Do you do Channel combination first or apply BXT to each narrowband channel . Also use PCC first or after. thanks
I using BlurXTerminator and it's a great tool!!! I see that you use FWHM value as PSF diameter, when I used that value calculated by PSFImage for my images I got a lot of artefacts. But I got a perfect result when I use r from PSFImage value as PSF diameter.
I've been impressed by the tool and the tips you've provided are very helpful. As a mono imager, I'm clear about running the tool separately on the luminance and combined RGB channels. For narrowband, do you run the tool separately for each channel and then combine?
That's my question as well. Typically I would create a lumincence of Ha, apply BlurXterminator and NoiseXterminator to the Ha Lum, and then overlay it (after stretching) to the combined greyscale SII, Ha, OIII channels (also stretched) - generally after combining those channels I would blur - convolution - them and that's all. So HaLum goes over those channels. Not sure though if this is the approach to follow with the BlurXterminator process. Russell do you have any guidance for us narrowband imagers??
This was an extremely helpful video, THANK YOU. I tried to download the psf web address that you posted. Unfortunately, no luck. Can you double check to verify that it is the correct address? Much thanks.
I am using a W.O. FLT91 with ASI6200MC. On stacks of (20) - (80) 300s subs, I am not seeing significant improvement in the image. I can generally obtain sharper results generating a high pass filter in Photoshop or processing it in Topaz Sharpen AI.
I've really been enjoying testing BXT. I'm getting the same results as you with auto PSF and manual PSF. It's subtle, but there's a difference. Only with one image did I not find any difference. I imagine that's because auto PSF and manual PSF agreed. That makes me wonder what auto PSF is measuring.
This is awesome! Need to try this out... have a question when applying the BlurXterminator you're saying to combine the images and then apply it or apply it to each filter? Thank you for the awesome video. Keep up the great work.
Hey Llamar - this is something I chatted with Russ about. At the moment, early on, there isn't necessarily a wrong approach with regards to combining and BXT or applying to each separate channel. I think though applying to the combined image works best. I'd really only apply it separately if I if I was dealing with say Ha and OIII channels to mix into RGB. Then I'd apply it separately to the Ha and OIII and as well to the combined RGB.
Great vid Shawn, I have one question for you. When it comes to LRGB and SHO, LRGB 1st Is there a benefit to run BXT on each channel (master) prior to combining the RGB or is it preferred after you do a combine? and the same with the SHO masters?
I've found best results running BXT on the RGB combine. For SHO you could combine the channels and run BXT as well. I run BXT separately on Ha if I am mixing it in with RGB.
Can I apply BlurXterminator in this sequence > WBPP then SPCC followed by DBE then BlurX ? Or should it be done after WBPP (in which case i'll lose my astrometric solution).
For clarification - you're suggesting to run BxT twice on LRGB images - once on the RGB combined image and again on the L image, before doing an LRGB combination? And for narrowband data, only on the L channel?
It's early on with BXT so we have to see how best to utilize it. For LRGB I think it may be that it needs to be applied to both lum image and the RGB combine, separately. There's benefit for doing both. And BXT should be done as early as possible in the workflow. I need to experiment more with BXT still. Depending on your workflow it may be possible that applying to the lum only will work but I haven't tried that yet.
I agree with you Shawn - BlurXterminator is absolutely must in a Pixinsight workflow. Any thoughts on whether NoiseXterminator should come before or after BXT? Great videos as always and looking forward to your next one. Happy Holidays and Clear Skies!
Hi David. BlurXterminator should be done right away. After crop and DBE. And before stretching. You can then move on to NoiseXterminator etc etc in your workflow.
Definitely a game changer. Image processing is getting easier and easier which is undoubtedly a good thing, yet I can't help but feel that my images are less of an accomplishment by using all of these AI-powered software.
I hear ya. I've actually thought about that as well. the AI and these new tools are a good thing, but there is something to be said about being able to do something like decon on your own without assistance. I think its good to know how to do it yourself regardless. Gives you better insight how it all works.
Thanks Shawn for that very timely video! One question: I've noticed that BXT can introduce sharpening artefacts (like wormlike structures familiar from other sharpening tools) when going with very high values for the Nonstellar Sharpening. With my data I therefore rarely go above 0.50 there, usually I stick to 0.25 or 0.33. What has your experience been with that?
This is true of any deconvolution tool. The effect is similar to running too many iterations of the classical deconvolution algorithms. Any tool needs to be wielded with finesse and judgment.
Hi Shawn, after using BXT for several days, I found out that BXT is surely a great tool, and I also found out that using BXT will destroy the color of the stars, which makes the stars' color more purple. I do BXT to each RGB chanel before RGB combination, but it doesn't work. Do you have any solution to this problem?
BXT should be applied to the RGB combine image. Not to the individual channels. Should be applied early on in workflow as well. After crop and DBE is done. I haven't experienced any star colour issues.
Are you sure these details are real or blurxterminator made a personal interpretation of them? interesting to compare this image with a Hubble telescope or on the same nebula taken with a 5m telescope.
Its a great tool as all of Russ's AI tools work great and he helps the hobby. The price is steep compared to StrEx, NoiseEx even with the multi discount its $90. It is a great tool.
BlurXTerminator doesn't create any detail that doesn't stem from information in your image: it works only with low-contrast (blurred) detail actually present in the image. It *can* create artifacts by over-sharpening this information if not wielded properly, but this is true of the classical deconvolution algorithms as well.
@@rrcroman, similar question.. after seeing the comparison at 5:55, I too was suspicious that your tool might be synthesizing data in the same way that Topaz AI does, but it sounds like your use of AI is for configuring deconvolution rather than pattern matching and insertion like Topaz?
Correct. Neural nets can be trained to be "generative," which is to say they create something out of nothing. I don't know specifically how Topaz DN was trained, but to my eye it looks generative. Neural nets can also be trained to approximate mathematical functions like deconvolution with high accuracy. The allowable training methods to achieve this are very restricted, and all generative training methods (GANs, diffusion, perceptual losses, etc., for those who understand machine learning) must be be strictly avoided. BlurXTerminator's network was trained this way.
Shawn... I enjoy learning from your videos and appreciate your clear explanations and instructions. But... I find the whoshing sound between segments here to be distracting and sort of annoying. Don't go down the rabbit hole that caught Trevor Jones. His videos started out nicely, but he has gotten into way over-producing. That and his BS about the OPT Radian Raptor just became too much. And, yes, I am also a fan of Russ's work and am on a trial of BlurX. ...Stephen
Thanks for the input/feedback. I like the whooshing! :) lol. Just trying to make the edits a little more interesting. Maybe going down the rabbit hole like Trevor would be good - he has a lot of subscribers.
Thanks Shawn. In is documentation , Russel write : "if Automatic PSF is selected, any aberration correction (coma, astigmatism, etc.) applied to stars will also be applied to nonstellar features. This is not true in manual mode (unless the Correct First option is selected):the PSF of nonstellar features will be assumed to be round."
So maybe, it will be beter to select "correct First" option when you use manual mode?
Sorry for my english, i'm french..😉
Yea for sure brought it as soon as I saw Adam's vid its so great to go over old data
Thanks!
Thank you! Appreciate the support.
Since I had completely avoided convolution completely, there was no struggle. ;-) BXT impact is something that may or may not be that noticeable at "screen size" viewing. But its impact becomes more apparent as the image size increases. In my M45 that was posted earlier this week, I applied BXT as I was just about through with the image. When zoomed in the increased detail around the primary stars and their shock waves in the dust cloud have noticeable improvements in detail/contrast.
Great work Shawn! - The tool itself has invigorated me for astronomy even more than I already was, a true game changer I reckon!
The PSF script looks really handy by the way, thanks for the tip!! 🙂
Clear skies buddy, hope you and yours have a lovely christmas and new year!
Thanks Luke. Appreciate you taking the time to watch and comment. I saw your vid as well. Nicely done. Happy Christmas to you and yours!
Excellent Shawn, I wonder if the star reduction makes it more difficult for StarNet which is still a big part of my process.
Actually I noticed an improvement! (at least on StarExterminator)
@@Jcastleinfo Good to know, thanks
@@Jcastleinfo I've had mixed results with StarXT - sometimes it throws weird artifacts when using after BlurXT.
I bought it the minute I saw it a few days ago. I have been having a reprocessing marathon and the results are nothing short of amazing. Thanks for this video as everything you guys post is appreciated.
Fantastic, what a result. Thanks Shawn, another great presentation. Having struggled with deconvolution for too long this is going to be my next purchase.
Great stuff! Bought it as soon as I saw it. It's funny that I was just recently wondering if Russ could do something with Deconvolution! I had mostly given up despite your great tutorial. Thanks also for the tip on PSFimage too. I did not have that one yet. The few images I tried so far I used this after SPCC but will move it up to post DBE now. Cheers!
It’s a no-brainier and a game-changer… I’m wanting to frame my images zoomed in more now I have all this extra detail. I’ve done a few re-edits and the one thing I have discovered is that it’s easy to over-process now… BlurX does such an awesome job that I think my workflow will be a lot less after combining than it was prior… it’s amazing 👍🏻
Sean,
Just started my processing after WBPP of M42 (LSHO). I set up the Ha image with two previews. On the first I ran BXT at defaults, then I ran BXT with the PSF determined by PSF analyzer. I as blown away by the difference, especially in cloud details. It is definitely worth running the PSF analysis and taking that chore off of the BXT. I have to say that the RUclips videos really don't do justice to the impact of BXT, especially with the PSF 'tuned' in.
Thanks for the video. I am just now trying the BlurXterminator on the heart nebula with fantastic results. One thing that I notice is that the process is quite slower than what I saw in your video. I have a really poweful computer game, top cpu, memory (32GB), etc; and still we take a good time to complete.
I have the Nvidia hack done so that blurX and noiseX use the GPU to perform their tasks. Not the CPU which is slower. It's a bit involved to do the hack but it does work well.
@@VisibledarkAstro thanks for replying it. I tried to do the Nvidia hack but it didn't work. Do you have a video explaining how to do it?
@@alexalmeida8627 Did you follow William Li's method? There was also one additional file that needed to be installed (according to Russ) to make it work. This was for StarX but may also help with speeding up BXT. If you get in touch with me via email (see my website visibledark.ca for contact), I can send you the links.
Your videos have helped me more than anyone else on RUclips. Thank you from a grateful subsy
Gosh Darn it Shawn! I was trying to avoid watching this video but finally broke down, your results speak for for themselves. I have Russ's other products and figured I would probably have to have this one as well. Thanks for another excellent PixInsight tutorial! Cheers Kurt
Thank you Shawn... great down to earth instructions
Was reading the document for Blurxterminator and noticed a comment about diffraction spikes that I get with my reflector. Not sure if it was saying it would correct for those? In some pictures, I like the dramatic affect of the spikes. Definitely going to purchase this. And loved all the comments about workflow.
I use a reflector as well. Doesn't affect the diffraction spikes that I've seen. Great tool to have in the workflow!
Good to know. Thanks!
I got trial version few days ago.Tool is absolutely amazing.No more straggling with deconvolution 🙂.Thanks for sharing tips how to further improve image
So funny. I went to Cloudy Nights this weekend asking for some help reducing star size with open star clusters and someone recommended this. I took one look at it and just bought it. Like fantastic.
Thanks Shawn, I just made the purchase. As an aside note, if you pay with PayPal make sure of the email access that you are using for Pay Pal.
Great tutorial - and looking very trim too Shawn!
Thank you Shawn for another great video. I experimented BXT and it save a lot of time in processing and the image look great after.
Glad it was helpful!
I've never used deconvolution, maybe now I'll do it with this program. It always seem to make a little difference for a lot of trouble when I tried it.
Give it a try. Great results with BXT. And makes decon a lot easier. I actually wouldn't do decon (the old way) for some of my images because it was to much work! lol. Now BXT changes all that.
Great tool! I downloaded the trial and tried it on a few of my images with excellent results. I'm definitely purchasing it!
I just signed up for the free trial! thanks for making this video!
Great video thanks Shawn. When you blink between before and after comparisons, what keyboard key/shortcut are you using please?
Hey Julian - sorry for the delay. I get swamped with questions and it is difficult to answer all of them and timely. I press control-shift and Z to do it. It can only be done on previews.
@@VisibledarkAstro thanks Shawn, no worries at all. I can understand the delay but please keep these videos coming, they are a great help.
So just to be clear this is best applied before stretching the image?
As with all of your video tutorials, excellent stuff, thanks
So far you have the best video on BXT . Couple of questions . Do you do Channel combination first or apply BXT to each narrowband channel . Also use PCC first or after. thanks
I channel combine the RGB first. Crop, DBE, SPCC and then do BlurX. That would be a correct order to do things.
I using BlurXTerminator and it's a great tool!!! I see that you use FWHM value as PSF diameter, when I used that value calculated by PSFImage for my images I got a lot of artefacts. But I got a perfect result when I use r from PSFImage value as PSF diameter.
I've been impressed by the tool and the tips you've provided are very helpful. As a mono imager, I'm clear about running the tool separately on the luminance and combined RGB channels. For narrowband, do you run the tool separately for each channel and then combine?
That's my question as well. Typically I would create a lumincence of Ha, apply BlurXterminator and NoiseXterminator to the Ha Lum, and then overlay it (after stretching) to the combined greyscale SII, Ha, OIII channels (also stretched) - generally after combining those channels I would blur - convolution - them and that's all. So HaLum goes over those channels. Not sure though if this is the approach to follow with the BlurXterminator process. Russell do you have any guidance for us narrowband imagers??
What are the option check boxes for in under all of the sliders?
This was an extremely helpful video, THANK YOU. I tried to download the psf web address that you posted. Unfortunately, no luck. Can you double check to verify that it is the correct address? Much thanks.
I am using a W.O. FLT91 with ASI6200MC. On stacks of (20) - (80) 300s subs, I am not seeing significant improvement in the image. I can generally obtain sharper results generating a high pass filter in Photoshop or processing it in Topaz Sharpen AI.
Interesting. Thanks for comment. I've seen great results with it. Did you manually determine and adjust your PSF setting?
Excellent. I purchased it and it is a game changer
I've really been enjoying testing BXT. I'm getting the same results as you with auto PSF and manual PSF. It's subtle, but there's a difference. Only with one image did I not find any difference. I imagine that's because auto PSF and manual PSF agreed. That makes me wonder what auto PSF is measuring.
This is awesome! Need to try this out... have a question when applying the BlurXterminator you're saying to combine the images and then apply it or apply it to each filter? Thank you for the awesome video. Keep up the great work.
Hey Llamar - this is something I chatted with Russ about. At the moment, early on, there isn't necessarily a wrong approach with regards to combining and BXT or applying to each separate channel. I think though applying to the combined image works best. I'd really only apply it separately if I if I was dealing with say Ha and OIII channels to mix into RGB. Then I'd apply it separately to the Ha and OIII and as well to the combined RGB.
Great vid Shawn, I have one question for you. When it comes to LRGB and SHO, LRGB 1st Is there a benefit to run BXT on each channel (master) prior to combining the RGB or is it preferred after you do a combine? and the same with the SHO masters?
I've found best results running BXT on the RGB combine. For SHO you could combine the channels and run BXT as well. I run BXT separately on Ha if I am mixing it in with RGB.
Thank you , Shawn!
good simple explanation. great tutorial
Can I apply BlurXterminator in this sequence > WBPP then SPCC followed by DBE then BlurX ? Or should it be done after WBPP (in which case i'll lose my astrometric solution).
You can do it that way yes. Most important is to apply it in the linear state and as early as possible. Your results should be fine.
For clarification - you're suggesting to run BxT twice on LRGB images - once on the RGB combined image and again on the L image, before doing an LRGB combination? And for narrowband data, only on the L channel?
It's early on with BXT so we have to see how best to utilize it. For LRGB I think it may be that it needs to be applied to both lum image and the RGB combine, separately. There's benefit for doing both. And BXT should be done as early as possible in the workflow. I need to experiment more with BXT still. Depending on your workflow it may be possible that applying to the lum only will work but I haven't tried that yet.
Hi. Im using PSF image and my value is over 8, what did i wrong ?
I agree with you Shawn - BlurXterminator is absolutely must in a Pixinsight workflow. Any thoughts on whether NoiseXterminator should come before or after BXT? Great videos as always and looking forward to your next one. Happy Holidays and Clear Skies!
Hi David. BlurXterminator should be done right away. After crop and DBE. And before stretching. You can then move on to NoiseXterminator etc etc in your workflow.
Make it available for Photoshop please.
Definitely a game changer. Image processing is getting easier and easier which is undoubtedly a good thing, yet I can't help but feel that my images are less of an accomplishment by using all of these AI-powered software.
I hear ya. I've actually thought about that as well. the AI and these new tools are a good thing, but there is something to be said about being able to do something like decon on your own without assistance. I think its good to know how to do it yourself regardless. Gives you better insight how it all works.
Thanks Shawn for that very timely video! One question: I've noticed that BXT can introduce sharpening artefacts (like wormlike structures familiar from other sharpening tools) when going with very high values for the Nonstellar Sharpening. With my data I therefore rarely go above 0.50 there, usually I stick to 0.25 or 0.33. What has your experience been with that?
This is true of any deconvolution tool. The effect is similar to running too many iterations of the classical deconvolution algorithms. Any tool needs to be wielded with finesse and judgment.
Thanks Russell for your answer! Finesse it will be then 😁
Hi Shawn, after using BXT for several days, I found out that BXT is surely a great tool, and I also found out that using BXT will destroy the color of the stars, which makes the stars' color more purple. I do BXT to each RGB chanel before RGB combination, but it doesn't work. Do you have any solution to this problem?
BXT should be applied to the RGB combine image. Not to the individual channels. Should be applied early on in workflow as well. After crop and DBE is done. I haven't experienced any star colour issues.
Wow! Amazing for sure. I bought it. No brainer.
Amazing results🤩
Are you sure these details are real or blurxterminator made a personal interpretation of them?
interesting to compare this image with a Hubble telescope or on the same nebula taken with a 5m telescope.
Its a great tool as all of Russ's AI tools work great and he helps the hobby. The price is steep compared to StrEx, NoiseEx even with the multi discount its $90. It is a great tool.
Another great video
Yea.. this tool is amazing! Yet again Russel is going to force us to reprocess a bunch of older data, lol. CS Shawn!
So true! I've already been reprocessing old data. lol.
how about the ethical site of this tool. how real are the images. i am not a good photographer. but when using this tool. i look like a pro!
BlurXTerminator doesn't create any detail that doesn't stem from information in your image: it works only with low-contrast (blurred) detail actually present in the image. It *can* create artifacts by over-sharpening this information if not wielded properly, but this is true of the classical deconvolution algorithms as well.
@@rrcroman, similar question.. after seeing the comparison at 5:55, I too was suspicious that your tool might be synthesizing data in the same way that Topaz AI does, but it sounds like your use of AI is for configuring deconvolution rather than pattern matching and insertion like Topaz?
Correct. Neural nets can be trained to be "generative," which is to say they create something out of nothing. I don't know specifically how Topaz DN was trained, but to my eye it looks generative. Neural nets can also be trained to approximate mathematical functions like deconvolution with high accuracy. The allowable training methods to achieve this are very restricted, and all generative training methods (GANs, diffusion, perceptual losses, etc., for those who understand machine learning) must be be strictly avoided. BlurXTerminator's network was trained this way.
@@rrcroman thank you for confirming. Purchased!
@@rrcroman Can you demostrate this?
Now I have to go redo all my images! Lol
russ, please make it working for photoshop!!😂
no works on linear data, is writen on site
Is it cheating? I lost my respect Bart delsaert he did it on one of his images
someone should tell nasa about BlurXTerminator
They will not need to explore space in the future....they will ask at AI inside an office.
Shawn... I enjoy learning from your videos and appreciate your clear explanations and instructions. But... I find the whoshing sound between segments here to be distracting and sort of annoying. Don't go down the rabbit hole that caught Trevor Jones. His videos started out nicely, but he has gotten into way over-producing. That and his BS about the OPT Radian Raptor just became too much. And, yes, I am also a fan of Russ's work and am on a trial of BlurX. ...Stephen
Thanks for the input/feedback. I like the whooshing! :) lol. Just trying to make the edits a little more interesting. Maybe going down the rabbit hole like Trevor would be good - he has a lot of subscribers.
AI is a delusion.