BlurXTerminator - Wow!

[Stuart1971]

2 hours ago, powerlord said:

I've been trying it on more images and yes, galaxies it seems to understand and do a good job. Better than other methods I'd been using ? Difficult to say without taking each through a new processing cycle. I mean, yeh it clearly 'understands M51' and you get lots of lovely detail. But I seem to have managed to extract most of that detail in other ways when I look at the final image I arrived at. So - it works.

So showing before and afters from raw to after blurX is interesting, but not a fair comparison really - It needs more 'doing whatever you did before blurx to get to a final image vs now with blurX in you processing workflow' really.

Just a pity it's PI only.

Nebulas - not a sausage though 😞

 

So seems pretty clear cut to me it's ony useful for galaxies  - star shapes aside. I won't be buying it until its available as an affinity plugin. come on Russell - there's more to Life than Pi (sic)

 

 

Can I ask, why are you so anti PI, it really is an incredible piece of software, and not that hard to use either, ok, so nowhere near as cheap as Affinity, but when there is a new affinity version you have to buy it, so in time you will end up spending as much….Just curious….🤔

[teoria_del_big_bang]

2 hours ago, powerlord said:

I've been trying it on more images and yes, galaxies it seems to understand and do a good job. Better than other methods I'd been using ? Difficult to say without taking each through a new processing cycle. I mean, yeh it clearly 'understands M51' and you get lots of lovely detail. But I seem to have managed to extract most of that detail in other ways when I look at the final image I arrived at. So - it works.

So showing before and afters from raw to after blurX is interesting, but not a fair comparison really - It needs more 'doing whatever you did before blurx to get to a final image vs now with blurX in you processing workflow' really.

Just a pity it's PI only.

Nebulas - not a sausage though 😞

So seems pretty clear cut to me it's ony useful for galaxies  - star shapes aside. I won't be buying it until its available as an affinity plugin. come on Russell - there's more to Life than Pi (sic)

I just have not had enough time to evaluate it enough yet to give a firm opinion, but what I have tried so far and from what I have seen on Adams video is that as much as anything it is the ease of use that is what is so desirable. 
To get great images some form of convolution is usually necessary, if only in the Luminance, and for me at least in PI this is one process I have found extremely tedious and time consuming, to get good results without noticeable artefacts, whereas in BlurXT it is do easy and so fast to perform, even if you need to undo and try different slider options a few times it still does not take long, or require any major effort.
And at the same time the stars get tightened (quite dramatically if you set the sliders to do so) and halos can be reduced.
Even if the plan is to go starless and process the stars separately, which seems to be a fairly common processing method these days,  I think reducing them and reducing halos can give better starless images and so maybe worth doing this before StarXT, or whatever you use.

And yes it does seem to work really well on Galaxies but again from what I have seen so far, rather than what I have produced, it does also seem to work well on nebulas with lots of filamentary detail, but I think you could be right that on some less filamentary nebulas the results are not so spectacular and maybe even marginal.

Unfortunately, for non PI owners, or those who do not take to it, unless it does come out as a stand alone or plug in then maybe that does not make it a viable option as purchasing both PI and BlurXT is expensive to basically just use one process.

Steve

[powerlord]

1 hour ago, Stuart1971 said:

Can I ask, why are you so anti PI, it really is an incredible piece of software, and not that hard to use either, ok, so nowhere near as cheap as Affinity, but when there is a new affinity version you have to buy it, so in time you will end up spending as much….Just curious….🤔

Acht I'm not sure this is the topic for that Stuart. It's not about cost. But you are right, I am anti pi. I make no secret of that. Each to their own of course.

Having just spent several hours putting all my galaxy stacks through blurx trial I admit its amazing for galaxy stuff. But nothing about what it does is because of PI. I use noiseX and starX fine in affinity. Why on earth would I change my whole processing work flow to be PI based to be able to use blurx for galaxies? 

I'm oooo 3 years in now (wow!), and believe me I've evaluated pi, done tutorials, etc. And I'm happy I can get as good or better results with APP, siril and affinity, plus Russell's plugins. If Russell doesn't see sense, and the possibility of extra revenue from a standard plugin, so be it. I'll do without. 

[Stuart1971]

26 minutes ago, powerlord said:

Acht I'm not sure this is the topic for that Stuart. It's not about cost. But you are right, I am anti pi. I make no secret of that. Each to their own of course.

Having just spent several hours putting all my galaxy stacks through blurx trial I admit its amazing for galaxy stuff. But nothing about what it does is because of PI. I use noiseX and starX fine in affinity. Why on earth would I change my whole processing work flow to be PI based to be able to use blurx for galaxies? 

I'm oooo 3 years in now (wow!), and believe me I've evaluated pi, done tutorials, etc. And I'm happy I can get as good or better results with APP, siril and affinity, plus Russell's plugins. If Russell doesn't see sense, and the possibility of extra revenue from a standard plugin, so be it. I'll do without. 

So how are you now using the BlurX on your galaxies…as you don’t use PI…🤔

[david_taurus83]

9 hours ago, powerlord said:

I've found the same thing - yes, it does stars well, but does nowt at all with nebulosity on anything I've tried on it. 

I thought the same with my M45 I processed yesterday but that's quite coarse pixel scale at over 3" per pixel. I tried with same Ha data I had already processed on the Tulip Nebula and yet to process Tadpoles. Its subtle but it does work. It even sharpens the dark dust so I don't think it's to do with SNR but more so on high contrast areas. These gifs probably aren't the best quality to showcase the changes, it looks better in PI but you get the idea. It's a very good deconvolution tool, much better than the PI deconvolution process. I've never gotten anything as good out of that. These gifs are BlurX applied directly to the native stacks of Ha data on each nebula.

 

[tomato]

Spending £300+ on software to use only one or two tools does sound excessive but having just gone for an IMX571 mono camera for £2K, I think that wrt image improvement the software option will give me more bangs for the buck compared to the hardware outlay.

[powerlord]

45 minutes ago, Stuart1971 said:

So how are you now using the BlurX on your galaxies…as you don’t use PI…🤔

I'm trialing BlurX in a trial of PI to see what all the fuss is about!

[powerlord]

12 minutes ago, david_taurus83 said:

I thought the same with my M45 I processed yesterday but that's quite coarse pixel scale at over 3" per pixel. I tried with same Ha data I had already processed on the Tulip Nebula and yet to process Tadpoles. Its subtle but it does work. It even sharpens the dark dust so I don't think it's to do with SNR but more so on high contrast areas. These gifs probably aren't the best quality to showcase the changes, it looks better in PI but you get the idea. It's a very good deconvolution tool, much better than the PI deconvolution process. I've never gotten anything as good out of that. These gifs are BlurX applied directly to the native stacks of Ha data on each nebula.

 

that is good. I did nowt with my tadpoles. The only nebula so far trialing it that it's done anything on is the flame and horse head nebula. weird.

[geeklee]

24 minutes ago, david_taurus83 said:

These gifs probably aren't the best quality to showcase the changes, it looks better in PI but you get the idea. It's a very good deconvolution tool, much better than the PI deconvolution process. I've never gotten anything as good out of that. These gifs are BlurX applied directly to the native stacks of Ha data on each nebula.

Very clear examples David, nice 👍  I'm seeing similar on RedCat data at 3.8"PP.  Subtle on larger structures but more obvious on small scale stuff (still subtle though)

24 minutes ago, david_taurus83 said:

Tadpoles

These go from being good to real stand outs after BlurX.  That little bit of deconvolution done by BlurX really makes them shine.

Here's the Crescent from my RedCat so 3.8"PP.  Left, original Ha stack.  Middle with "Nonstellar" at 0.6 then right at 0.9

My finished image was close but BlurX would have added a touch extra.

I see subtle improvements in all the nebulosity stacks I've looked at and that's just the undersampled raw Ha.

Edited December 20, 2022 by geeklee

[teoria_del_big_bang]

14 minutes ago, david_taurus83 said:

I thought the same with my M45 I processed yesterday but that's quite coarse pixel scale at over 3" per pixel. I tried with same Ha data I had already processed on the Tulip Nebula and yet to process Tadpoles. Its subtle but it does work. It even sharpens the dark dust so I don't think it's to do with SNR but more so on high contrast areas. These gifs probably aren't the best quality to showcase the changes, it looks better in PI but you get the idea. It's a very good deconvolution tool, much better than the PI deconvolution process. I've never gotten anything as good out of that. These gifs are BlurX applied directly to the native stacks of Ha data on each nebula.

Thanks for taking time to post these, they are great demonstrations and I don't think they are so subtle, After all the ide is sharpening what is actually on our images and not to add any significant artefacts so in most images the results will be very much like you have demonstrated here.
As I said before I always struggled with deconvolution processes in PI before this, it always seemed so difficult and what would work on one image would not work directly on another image so I always spent ages twiddling to get a half decent result.

Steve

[tomato]

My experience to date is I see more noticeable improvements on 'deep' data sets, maybe the AI has more signal to work with? Here are NGC 7331 crops (20 hrs of OSC data), one with SPCC and BlurXterminator (default settings).

 

30 minutes ago, powerlord said:

I'm trialing BlurX in a trial of PI to see what all the fuss is about!

Wow, you better put the coffee on!😄

[ngc1535]

11 hours ago, Lee_P said:

@ngc1535 I'd love to get your thoughts on why I'm not seeing much improvement using BlurX on the nebulosity in these stacks -- is it because I need a higher SNR, as was suggested earlier in this thread? Thanks in advance if you're able to help!

Correct. You do not have the S/N (in most areas) necessary to enjoy the benefits of deconvolution (of any type). 

-adam

[ngc1535]

1 minute ago, tomato said:

My experience to date is I see more noticeable improvements on 'deep' data sets, maybe the AI has more signal to work with? Here are NGC 7331 crops (20 hrs of OSC data), one with SPCC and BlurXterminator (default settings).

 

Wow, you better put the coffee on!😄

Regarding the information to work on (Tomato)- this is a BXT specific or special requirement.. .it has always been true for any deconvolution. 

[tomato]

2 minutes ago, ngc1535 said:

Regarding the information to work on (Tomato)- this is a BXT specific or special requirement.. .it has always been true for any deconvolution. 

I confess that until now I have not been a serious user of deconvolution tools, my attempts to use them have yielded little or no improvement. However, BXT is so simple and quick to use I can quickly establish the benefit or otherwise of it's application. It's great!

[powerlord]

Startools always worked well for me for galaxies. I just didn't like what it did with stars. Well, or anything else tbh.

[Stuart1971]

2 hours ago, david_taurus83 said:

I thought the same with my M45 I processed yesterday but that's quite coarse pixel scale at over 3" per pixel. I tried with same Ha data I had already processed on the Tulip Nebula and yet to process Tadpoles. Its subtle but it does work. It even sharpens the dark dust so I don't think it's to do with SNR but more so on high contrast areas. These gifs probably aren't the best quality to showcase the changes, it looks better in PI but you get the idea. It's a very good deconvolution tool, much better than the PI deconvolution process. I've never gotten anything as good out of that. These gifs are BlurX applied directly to the native stacks of Ha data on each nebula.

 

It works even better if you specify the PSF to run, use the tool in PI to see the FWHM figures for the image, and take the average and use that, it works even better then….rather than the auto option….👍🏻

[Stuart1971]

Just been doing some more messing around 

This was an unguided stack of Veil images I took with an old CCD, some time ago, I was testing my EQ8 unguided, these are 5 min unguided subs, and PA was about 4 mins off, there are about 20 images in the stack, see the top to bottom oblong stars from lack of guiding and poor PA, then ran the image through BlurX, and it’s almost a useable image, with more processing of course….

Before

 

After

Edited December 20, 2022 by Stuart1971

[Xiga]

Stuart, that's simply amazing how BXT has transformed that data! 

A lot of very skilled imagers have been happily making jaw-dropping images for years now without using any form of deconvolution (mainly because of it's dark art form, I guess). So if this now makes it a trivial affair, then that's great! For me though, the real wow factor comes from how it can correct star shapes. That just blows my mind tbh, and your images are a perfect example of what's possible now with this software. I'm also starting to wonder whether we still need all these expensive flatteners anymore, lol 😋

Ps - I read a few comments where people where saying they weren't seeing much improvement in non-stellar details. I think this could be down to not using the Manual PSF option. Here's a video that describes it pretty well. You can skip to 10:30 for that specific bit

 

Edited December 23, 2022 by Xiga

[Stuart1971]

Well, if you think BlurX was good, he has gone and done it again with this new tool….😂

 

[teoria_del_big_bang]

6 minutes ago, Stuart1971 said:

Well, if you think BlurX was good, he has gone and done it again with this new tool….😂

 

With the Yorkshire weather this year I think this might be my best way to go 🤣

Steve

[Lee_P]

12 hours ago, Xiga said:

Ps - I read a few comments where people where saying they weren't seeing much improvement in non-stellar details. I think this could be down to not using the Manual PSF option. Here's a video that describes it pretty well. You can skip to 10:30 for that specific bit

Thank you so much, I wasn't getting good results but after dialing in the Manual PSF it's making a big difference. @powerlord, if you, like me, were just using the automatic settings, then I recommend giving this approach a go. It's outlined in this video from 7:20.

 

 

[powerlord]

4 hours ago, Lee_P said:

Thank you so much, I wasn't getting good results but after dialing in the Manual PSF it's making a big difference. @powerlord, if you, like me, were just using the automatic settings, then I recommend giving this approach a go. It's outlined in this video from 7:20.

 

 

thanks, i did try that.

ive since been chucking lots more images at it. nebula wise, seems pretty hit and miss, but nothing like the difference i see on galaxies.

itll be interesting trying it on say galaxy images with my c925 without 6.3 reducer when i take some... see what magic it can do there

[teoria_del_big_bang]

A couple of related questions for those who use only Pixinsight for processing their images (or mostly PI anyway).

Regarding sharpening, do you think BlurXTerminator is all that is needed ? 
OR, might you still need further sharpening maybe at later, non-linear, stages that will still require one of the older processes previously used for this ?

Regarding noise reduction, do you think NoiseXTerminator is enough for this process?
OR, again do you think some of the other processes traditionally used for noise reduction offers something you cannot achieve with NoiseXT ?

I am not so experienced in AP, I also do not like to do too much in the way of noise reduction, I also have a pretty well behaved camera and to me I would answer yes to both of these questions but just wonder if I am missing something due to lack of experience. 
So, for me, NoiseXT seems to suffice and at least for the RGB side BlurXT seems all I need, maybe for a Luminance further processing to enhance the contrast is needed before adding to the RGB, that requires other processes ? 

For me I just would like to simplify my processing without missing an obvious step and some PI processes can get very tedious and complicated so using just these processes would be a Godsend for me.

Steve

[Lee_P]

25 minutes ago, teoria_del_big_bang said:

A couple of related questions for those who use only Pixinsight for processing their images (or mostly PI anyway).

Regarding sharpening, do you think BlurXTerminator is all that is needed ? 
OR, might you still need further sharpening maybe at later, non-linear, stages that will still require one of the older processes previously used for this ?

Regarding noise reduction, do you think NoiseXTerminator is enough for this process?
OR, again do you think some of the other processes traditionally used for noise reduction offers something you cannot achieve with NoiseXT ?

I am not so experienced in AP, I also do not like to do too much in the way of noise reduction, I also have a pretty well behaved camera and to me I would answer yes to both of these questions but just wonder if I am missing something due to lack of experience. 
So, for me, NoiseXT seems to suffice and at least for the RGB side BlurXT seems all I need, maybe for a Luminance further processing to enhance the contrast is needed before adding to the RGB, that requires other processes ? 

For me I just would like to simplify my processing without missing an obvious step and some PI processes can get very tedious and complicated so using just these processes would be a Godsend for me.

Steve

I'm far from a PI expert so take other people's thoughts over mine, but for what it's worth I'm currently reprocessing a load of old data using these new tools. I find that I'm still often doing a little bit of extra sharpening in the non-linear stage, but NoiseXTerminator is all I'm using for noise reduction. I'm OSC, so different techniques may be better for mono.

[Stuart1971]

32 minutes ago, teoria_del_big_bang said:

A couple of related questions for those who use only Pixinsight for processing their images (or mostly PI anyway).

Regarding sharpening, do you think BlurXTerminator is all that is needed ? 
OR, might you still need further sharpening maybe at later, non-linear, stages that will still require one of the older processes previously used for this ?

Regarding noise reduction, do you think NoiseXTerminator is enough for this process?
OR, again do you think some of the other processes traditionally used for noise reduction offers something you cannot achieve with NoiseXT ?

I am not so experienced in AP, I also do not like to do too much in the way of noise reduction, I also have a pretty well behaved camera and to me I would answer yes to both of these questions but just wonder if I am missing something due to lack of experience. 
So, for me, NoiseXT seems to suffice and at least for the RGB side BlurXT seems all I need, maybe for a Luminance further processing to enhance the contrast is needed before adding to the RGB, that requires other processes ? 

For me I just would like to simplify my processing without missing an obvious step and some PI processes can get very tedious and complicated so using just these processes would be a Godsend for me.

Steve

I’m also far from an expert, but NoiseX I think is more than enough for controlling the noise…

Now as for BlurX on sharpening, well I have not even got around to even trying Deconvolution yet in my PI journey, so again BlurX seems enough for me, especially as it pretty much solved my headache i was getting from not being able to get perfectly round stars in one corner of my images….that was a massive plus for me….👍🏻