Thanks for the review! This is great feedback! I don't know if Astrosharp will ever be able to close the gap with BXT (Russell Croman does a great job at what he does), but that doesn't mean I won't try :) About the ringing: I think it has to do with the fact that Hubble images (what it is trained on) have rings around the stars due to the Airy disc diffraction pattern. Hopefully, that can be corrected in a future update!
@@deepskydetail Haha, this reply gave me a good chuckle. I think opening a Google Drive where people can add their own data for you to use to train on will speed up things.
Cuiv! Just wanted you to know that sometimes I play your channel while doing other things just to hear you in the background because your energy and enthusiasm makes me happy!
We should always support all open source/ free software. That usually tells me, that someone is putting his skills into a project that he likes to do and his motives are just being interested in the hobby. Those projects usually end up turning out good and benefits of us all.
Nice comparison. BXT certainly looks to be far superior at present, but it's nice to see another free option for people to use. I love BXT, SXT and NXT, but together they cost almost as much as Pixinsight itself.
Then again, they are much much cheaper than a camera, a decent field flattener or other astroequipment. You might use the equipment, you will use the processing tools.
I saw AstroSharp before and wow the software has really come along! Always good to have competition especially open source to keep the paid software developers pushing the envelope 🙂 Would be nice if they added a PSF size feature to the AstroSharp software and that may help fine tune the sharpening since it is the PSF that determines the amount of deconvolution needed. Thanks for the video Cuiv, very informative and cool! Clear Skies!
Thank you! That is actually what I've been trying to work on at the moment to improve the tool! It takes a lot more training and data curation is more difficult.
Awesome video, dude! You hit all of the questions I had about this software. Thank you! One quick question…did you leave Star Sharpening on the default setting in BlurX?
I've been trying AstroSharp for a few days now Quiv, and must say that on some images it works good and on others absolutely not. Needs a lot of work to make it really something that can compare wit BlurX. Great comparison Quiv !
I processed my image of M51 in Siril and ran Astrosharp on it. The result is very good but I agree it does tend to oversharpen the stars and increase the noise in certain areas, maybe starnet could be run first and the stars replaced later. As I cant afford to buy PI and the various addons I really hope Astrosharp continues to improve could be a game changer for us cash strapped astrophotographers
I suspect the blurexterminator uses a spatially-variant deconvolution PSP, which would be able to correct astigmatism and other variable PSP which change over the field. Hoping to see evolution of AstroSharp though ❤
Thanks again for addressing what I hoped I would soon see, this exact comparison. The comparisons are very telling, but I noticed one difference that didn't come up - the difference in contrast. It seemed evident in all of the examples that AstroSharp left the blacks richer. That look, positive to me, outweighs the attributes that became apparent on the stars and noise, relatively speaking. You touched on that slightly by noting that it increased the stars volume outside of the objects. My question is was this contrast difference from AstroSharp and what do you think about 'space' being black in astrophotography?
It may simply be down to the stretching itself, since in one case the stretching is done prior to sharpening, and in the other case it is done after - so searching for a specific background level could have yielded different stretching
Had a friend of mine not twisted my arm buying into PixInsight and then the triple threat BX SX and NX, I'd be running over to the AS download. I am a fan of free/open source software (I still use Siril for stacking) and certainly endorse it for folks new to the hobby. But now that I've been indoctrinated into the PI fraternity, I do see its advantages--but only after I suffered a traumatic hazing. I do feel that the battle between paid vs open source is a good thing--it does push the developers on both sides to work harder.
if you set the Point Spread Function manually in BlurXterminator you can get even better results. I use DynamicPSF Process to determine what the setting shout be. infact Russell Croman recomends doing this over using the auto setting.
BXT has new AI training in the works from what I’m told by Adam and hearing Russell say it himself on a recent stream, it’s supposed to address aberrations to a higher degree so we’ll see… I already own BXT but it’s nice to have options…
As someone who doesn't use PixInsight, I'm really envious of BlurXTerminator. It's great to see an open-source alternative in development. I gave it a go and wasn't 100% pleased with the results but am keep to see it continue to improve. When StarNet++ was first released it was not as good as StarXTerminator but now they produce very comparable results. Hopefully we'll see the same sort of improvement with AstroSharp.
I personally find that starnet 2 never works for me. In my most recent image it even lead me to purchase StarX and then it worked perfectly. I held out because typically starnet 2 was good enough but
Nice comparison. Free stuff is always pleasant, so I am looking forward to seeing what the next iteration of astrosharp might do for us. It looks like the over sharpening of the stars are the major concern with astrosharp. Did you try to process a starless image using Starnet++ and just trying the AI algorithm on it? Maybe just then, adding the star mask. Just a thought
So after stacking, I take the image into Siril to do background extraction, run Starnet and then do my initial stretch, before taking into Photoshop to finish it off. So I was wondering, would I nee to run AstroSharp before I run Starnet or would it be better to run after I've done my stretching and processing in Photoshop, then run it through AstroSharp?
As long as the image is stretched and the colors haven't been manipulated yet (besides initial calibration) you should be fine - and as others mentioned it might be good to try on the image with stars removed!
@CuivTheLazyGeek ... I have a new Player One Poseidon OSC coming at the end of the month. I am shifting from modified DSLR cameras to cooled OSC cameras. I have watched many videos on the camera, and I think it suits my finicky weather, also my, bortle 5 sky's. Do you have any tips on the switch, from one to the other?
@@CuivTheLazyGeek I'll look into it more carefully then. I thought that this was only the shiny app that loads and executes the pre-trained models. Thanks!
I can't help but wonder if a starless image (for the galaxy) might give better results. I'm currently processing a 2x drizzled version of M101 and I'm curious to see the results of a few experiments.
I ran AstroSharp on a image of M101 and zoomed in on the centre i thought it was improved. Astrophotography is an expensive hobby snd it is a shame BXT and its brother's can not be used as standalone options for those of us that cant justify the cost of PI.
Kudos to @deepskydetail and thank you for the review, I’ll download and spread the word…. Very happy this program is being developed, I think the starnet idea is worthy to use with this as it gets developed and perhaps a final oass in topaz might help if noise us introduced… I’ll be testing it all for sure
Without magnification, these softwares look similar. Could you get review of Voyager versus NINA? Maybe you will switch to Voyager, since supposed to be higher versatility.
@@CuivTheLazyGeek It will be great to see which one is better for remote observatory Voyager or NINA. I think NINA with plug ins is good enough. But it will be great to know option of yours. Thanks
I wonder what would happen if you did a star extraction and then ran AstroSharp on the starless image. That might help with the ringing around the stars.
Regarding the *open source* claim. As far as I can see, the Git repo -doesn't seem to actually contain the -*-source-*- but the compiled -*-binaries.-*- This would make it freeware.- EDIT: Woopsie. My bad. Seems that the program is written in a language *R* I've never heard off and the repo is layed out differently than I've used to. This combined with the fact that the repo is littered with binary files led me to first believe that there was no actual source code. It appears I was wrong, but I don't know enough to say anything for certain at this point. With the actual *source code,* anyone could modify, improve and compile it. Also I would imagine making it Linux and Mac compatible wouldn't be too much of a stretch for the community. Anyway, I am exited for the new free software to some day challenge blurX.
Nice start. Awesome for free but it still needs a lot more training. Right now it oversharpens a lot noise and creates artifacts in the noise. Still cool though...
I tried Astro Sharp and found it still has a way to go. It created much noise, is slow and found I can do a better job manually using tools I have. It also fails to open sometimes and has to be forced to re-start. Using it on an already sharpened photo results in poor result and is not recommended by the creator still tried it as a test. Hope it develops in the future as it is a fine idea...
Man, I am about to be 400$ lighter. I have been processing images for a while now, but I want them to be that next level. I have not had success with any blurx eqivalents for free and imo, they introduce too much other abberations. Still on the fence with pi price but as far as i can tell there is nothing else out there that is comparable.
Since AstroSharp doesn't work on linear images, I guess the best use is to try it on finished images that are already stretched. At this point in time It may slightly improve detail, at the cost of some background noise. It's a try and see option, but I'm glad there is a free option. The Xterminator series of plugins are overpriced, like everything else in astrophotography...
Mac users have a strong programming community. I bet they will contribute soon with a mac compatible GUI and the installer to that open source project. All the other code is most likely fully reusable
BXT seems to be the clear winner for those pixel peepers with a fatter wallet (or deeper credit hole...). 👍 I wonder if astro sharp can be defined to not be so aggressive?
Yep, as I was rechecking my results, Astro sharp almost looked like my own manual decon attempts - I think BXT has mastered decon protection far better
If you want RUclips to promote your videos more, stick a rainbow flag on your scope or within visibility of your studio and mention a few times that you've gone to thinking about going trans... you'll be at the top of RUclips promotion algorithm...
Thanks for the review! This is great feedback! I don't know if Astrosharp will ever be able to close the gap with BXT (Russell Croman does a great job at what he does), but that doesn't mean I won't try :)
About the ringing: I think it has to do with the fact that Hubble images (what it is trained on) have rings around the stars due to the Airy disc diffraction pattern. Hopefully, that can be corrected in a future update!
Woohoo, the one and only! Makes sense on the ringing, you need some crap quality data to train on :D
@@CuivTheLazyGeek haha yeah!! Maybe I'll just start adding in my own data as part of the training :P
@@deepskydetail Haha, this reply gave me a good chuckle.
I think opening a Google Drive where people can add their own data for you to use to train on will speed up things.
@@BBROPHOTO great idea. I have plenty of crappy data to offer!
Cuiv! Just wanted you to know that sometimes I play your channel while doing other things just to hear you in the background because your energy and enthusiasm makes me happy!
That's awesome Phil, glad I can bring that energy 😊
We should always support all open source/ free software. That usually tells me, that someone is putting his skills into a project that he likes to do and his motives are just being interested in the hobby. Those projects usually end up turning out good and benefits of us all.
Nice comparison. BXT certainly looks to be far superior at present, but it's nice to see another free option for people to use. I love BXT, SXT and NXT, but together they cost almost as much as Pixinsight itself.
Exactly - and since I bought PI looong ago, for me SXT BXT and NXT cost more than PI :)
Then again, they are much much cheaper than a camera, a decent field flattener or other astroequipment. You might use the equipment, you will use the processing tools.
BXT is simply amazing. It is good to see that AstroSharp is out there now as a free alternative, especially for those not running PI.
Completely agree!
I saw AstroSharp before and wow the software has really come along! Always good to have competition especially open source to keep the paid software developers pushing the envelope 🙂 Would be nice if they added a PSF size feature to the AstroSharp software and that may help fine tune the sharpening since it is the PSF that determines the amount of deconvolution needed. Thanks for the video Cuiv, very informative and cool! Clear Skies!
Thank you! That is actually what I've been trying to work on at the moment to improve the tool! It takes a lot more training and data curation is more difficult.
Thanks Dave as always!
@@deepskydetail And do some training using images from BlurXTerminator ? (THANKS!!!!)
Nice comparison mate!! Very interesting to see the differences 👌
Thanks Luke;
Awesome video, dude! You hit all of the questions I had about this software. Thank you! One quick question…did you leave Star Sharpening on the default setting in BlurX?
Yes in this example BXT was on default settings, including star sharpening!
I've been trying AstroSharp for a few days now Quiv, and must say that on some images it works good and on others absolutely not. Needs a lot of work to make it really something that can compare wit BlurX. Great comparison Quiv !
I mainly EAA with Sharpcap and then have a dabble at refining the image later, free tools are always welcome. Thank Cuiv.
Thanks Cuiv for all those fantastic infos.
My pleasure!
I processed my image of M51 in Siril and ran Astrosharp on it. The result is very good but I agree it does tend to oversharpen the stars and increase the noise in certain areas, maybe starnet could be run first and the stars replaced later. As I cant afford to buy PI and the various addons I really hope Astrosharp continues to improve could be a game changer for us cash strapped astrophotographers
Wait and see I guess!
I suspect the blurexterminator uses a spatially-variant deconvolution PSP, which would be able to correct astigmatism and other variable PSP which change over the field. Hoping to see evolution of AstroSharp though ❤
We'll see how Astro sharp progresses!
Fab video as usual! Great to see more deconvolution options. What country are you from by the way?
Thank you! Originally from France!
Thanks again for addressing what I hoped I would soon see, this exact comparison.
The comparisons are very telling, but I noticed one difference that didn't come up - the difference in contrast.
It seemed evident in all of the examples that AstroSharp left the blacks richer. That look, positive to me, outweighs the attributes that became apparent on the stars and noise, relatively speaking.
You touched on that slightly by noting that it increased the stars volume outside of the objects.
My question is was this contrast difference from AstroSharp and what do you think about 'space' being black in astrophotography?
It may simply be down to the stretching itself, since in one case the stretching is done prior to sharpening, and in the other case it is done after - so searching for a specific background level could have yielded different stretching
Had a friend of mine not twisted my arm buying into PixInsight and then the triple threat BX SX and NX, I'd be running over to the AS download. I am a fan of free/open source software (I still use Siril for stacking) and certainly endorse it for folks new to the hobby. But now that I've been indoctrinated into the PI fraternity, I do see its advantages--but only after I suffered a traumatic hazing. I do feel that the battle between paid vs open source is a good thing--it does push the developers on both sides to work harder.
Completely agree, competition is a good thing!
I used astrosharp on an image of M97 taken with the Dwarf 2 and it did a nice job of sharpening.
Good to know, thank you!
if you set the Point Spread Function manually in BlurXterminator you can get even better results. I use DynamicPSF Process to determine what the setting shout be. infact Russell Croman recomends doing this over using the auto setting.
BXT has new AI training in the works from what I’m told by Adam and hearing Russell say it himself on a recent stream, it’s supposed to address aberrations to a higher degree so we’ll see… I already own BXT but it’s nice to have options…
Yep, also looking forward to further improvements in BXT!
could have tried using astrosharp on a starless image, then add the stars back in after? could help with the issues with stars with astrosharp
I love BTX, it will be interesting to see where the new tool goes
Does AstroSharp correct aberrations and imperfect guiding, or does it only do sharpening with circular symmetry?
As someone who doesn't use PixInsight, I'm really envious of BlurXTerminator. It's great to see an open-source alternative in development. I gave it a go and wasn't 100% pleased with the results but am keep to see it continue to improve.
When StarNet++ was first released it was not as good as StarXTerminator but now they produce very comparable results. Hopefully we'll see the same sort of improvement with AstroSharp.
I personally find that starnet 2 never works for me. In my most recent image it even lead me to purchase StarX and then it worked perfectly. I held out because typically starnet 2 was good enough but
Nice comparison. Free stuff is always pleasant, so I am looking forward to seeing what the next iteration of astrosharp might do for us. It looks like the over sharpening of the stars are the major concern with astrosharp. Did you try to process a starless image using Starnet++ and just trying the AI algorithm on it? Maybe just then, adding the star mask. Just a thought
I didn't try yet, it didn't occur to me haha!
So after stacking, I take the image into Siril to do background extraction, run Starnet and then do my initial stretch, before taking into Photoshop to finish it off. So I was wondering, would I nee to run AstroSharp before I run Starnet or would it be better to run after I've done my stretching and processing in Photoshop, then run it through AstroSharp?
As long as the image is stretched and the colors haven't been manipulated yet (besides initial calibration) you should be fine - and as others mentioned it might be good to try on the image with stars removed!
A Great in depth comparison . Thank you
Thank you!
Looks great, I’m in 👍
Let us know how it goes!
Definitely check it out! Thanks for sharing 👍🏻
Thanks Jeff!
Nice video on Blur software.... thanks
My pleasure!
@CuivTheLazyGeek ... I have a new Player One Poseidon OSC coming at the end of the month. I am shifting from modified DSLR cameras to cooled OSC cameras. I have watched many videos on the camera, and I think it suits my finicky weather, also my, bortle 5 sky's.
Do you have any tips on the switch, from one to the other?
So, do you convert back to xisf to finish processing? Sorry if this is a stupid question, I’m relatively new to this.
Just use the output tiff
Great job comparing the two tools.
Thank you!
Nice. Been following this one for a while
Thank you!
Unfortunately a no-go for me, since I use a Mac for editing :) Thanks for sharing!
You got me really excited with the "open source" part... sadly it is still closed source. Very impressive tool though.
It does look open source, just written in R - at least I can see code in the github
@@CuivTheLazyGeek I'll look into it more carefully then. I thought that this was only the shiny app that loads and executes the pre-trained models. Thanks!
Curious if blending the sharpened image with the original might net a better result, like EZ HDR does for HDRMT.
That's a great idea, something to try out!
Thanks!!! is fantastic!!!!
I can't help but wonder if a starless image (for the galaxy) might give better results. I'm currently processing a 2x drizzled version of M101 and I'm curious to see the results of a few experiments.
You're very likely right!
@@CuivTheLazyGeek I tried it and was happy with the result. I don’t have BlurXterminator to compare with.
What about applying astrosharp to a starless image then applying and reducing the stars afterwards?
Probably the way to go, yes!
It's a very good use for AI, not inventing like improving what a human and nature can create.
Sir plz try to make a video on zooming into the sunspots just before the sunset...
I ran AstroSharp on a image of M101 and zoomed in on the centre i thought it was improved. Astrophotography is an expensive hobby snd it is a shame BXT and its brother's can not be used as standalone options for those of us that cant justify the cost of PI.
Kudos to @deepskydetail and thank you for the review, I’ll download and spread the word…. Very happy this program is being developed, I think the starnet idea is worthy to use with this as it gets developed and perhaps a final oass in topaz might help if noise us introduced… I’ll be testing it all for sure
Definitely has potential
For sure!
Someone needs to add a feature to the software where you stack the two pictures and move a slider left and right to see A or B. And a quick blink.
Without magnification, these softwares look similar.
Could you get review of Voyager versus NINA? Maybe you will switch to Voyager, since supposed to be higher versatility.
I'll look into that - hard for me since I'm so used to NINA and it already does everything I need :-)
@@CuivTheLazyGeek It will be great to see which one is better for remote observatory Voyager or NINA. I think NINA with plug ins is good enough. But it will be great to know option of yours. Thanks
I wonder what would happen if you did a star extraction and then ran AstroSharp on the starless image. That might help with the ringing around the stars.
That's a great idea! More stuff to check!
Regarding the *open source* claim. As far as I can see, the Git repo -doesn't seem to actually contain the -*-source-*- but the compiled -*-binaries.-*- This would make it freeware.-
EDIT:
Woopsie. My bad. Seems that the program is written in a language *R* I've never heard off and the repo is layed out differently than I've used to. This combined with the fact that the repo is littered with binary files led me to first believe that there was no actual source code. It appears I was wrong, but I don't know enough to say anything for certain at this point.
With the actual *source code,* anyone could modify, improve and compile it. Also I would imagine making it Linux and Mac compatible wouldn't be too much of a stretch for the community.
Anyway, I am exited for the new free software to some day challenge blurX.
Thanks so much for checking! Yeah R takes me back to my university statistics courses :-)
Nice start. Awesome for free but it still needs a lot more training. Right now it oversharpens a lot noise and creates artifacts in the noise. Still cool though...
Exactly - it will be interesting to see how it evolves!
Why wouldn't you sharpen the starless and stars separately ? Wouldn't that make more sense ?
It would! I didn't think of it because I'm an idiot :D
I tried Astro Sharp and found it still has a way to go. It created much noise, is slow and found I can do a better job manually using tools I have. It also fails to open sometimes and has to be forced to re-start. Using it on an already sharpened photo results in poor result and is not recommended by the creator still tried it as a test. Hope it develops in the future as it is a fine idea...
Yep, but it's always good to see alternatives!
@@CuivTheLazyGeek I hope they improve on it in the future! It could turn into something pretty handy....
Man, I am about to be 400$ lighter. I have been processing images for a while now, but I want them to be that next level. I have not had success with any blurx eqivalents for free and imo, they introduce too much other abberations.
Still on the fence with pi price but as far as i can tell there is nothing else out there that is comparable.
All we need now is a free noise x terminator.
The GraXpert team is working on one :)
Since AstroSharp doesn't work on linear images, I guess the best use is to try it on finished images that are already stretched. At this point in time It may slightly improve detail, at the cost of some background noise. It's a try and see option, but I'm glad there is a free option. The Xterminator series of plugins are overpriced, like everything else in astrophotography...
Revolutionalized 😂😂😂😂😂😂
Hey I can complexifalize words as I like!!
Any Ai that can help with egg shaped stars on the edges?
Better to not care about them :)
@@CuivTheLazyGeek lol
I want a magic wand for all the pixel peepers ;-)
Mac users have a strong programming community. I bet they will contribute soon with a mac compatible GUI and the installer to that open source project. All the other code is most likely fully reusable
Yep just a matter of time
Me, as a Mac user, watching the first 4 minutes of this video: 🙂😀😃😐😕😞
I'm so sorry!!
@@CuivTheLazyGeek All good! As an astrophotography noob I’m loving the content, regardless.
Supernova in M101 Hurry to image it before it fades😊
Yesssssss but clouds lol
I've seen that for astrosharp it is better to do it before stretching
I'm waiting for a software that makes all the stars pinpoint, so I avoid spending money on correctors / flatteners lol
In a way BXT is close to that already lol
BXT seems to be the clear winner for those pixel peepers with a fatter wallet (or deeper credit hole...). 👍 I wonder if astro sharp can be defined to not be so aggressive?
Windows only:
Me: 🚶♂️ ✌️
Sorry... I'm sure it will come!
Do we lose something if sofware gets so good it makes the hobby to easy.
There's always the option to not use it ;)
It looks like astrosharp is doing deconvolution, while blurexterminator has other layers in a deep neural network or similar…
Yep, as I was rechecking my results, Astro sharp almost looked like my own manual decon attempts - I think BXT has mastered decon protection far better
Dunno. Not really keen on adding 32,000 files to my drive...
And quite a bit of drive space
Same here, too much noise.
Yep, it needs training with real noisy images
If you want RUclips to promote your videos more, stick a rainbow flag on your scope or within visibility of your studio and mention a few times that you've gone to thinking about going trans... you'll be at the top of RUclips promotion algorithm...