Newest modelling loaded. trained on a deeper neural net and added more data. I need to find a better way to telling the broader audience when a new version is available. Windows Version: drive.google.com/file/d/1WQvQvrrLFgzbZwe-exeHjECmawHKYGyS/view?usp=drive_link MacOS Version: drive.google.com/file/d/1Tze3E6b_si8ZjUXkPMj8TICqrP88ulbJ/view?usp=drive_link
Another great effort on your part Frank. I know you refer to this as the "golf cart" compared to the "gold standard", but I think this can only promote constructive competition amongst developers that in the long run will benefit everyone. Thanks.
You should shoot me an email, I know you are working a sharpening tool as well. I'm more than happy to collaborate and share my code on this! I think you are much more knowledgeable in this than I am. frank@setiastro.com
Still a beta, we can change it :) I did have it Seti Astro Sharpen, but that is too close to Deep Sky Detail's sharpen tool he is working on AstroSharp
You're amazing, thank you. While the RC astro suite is outstanding. The price is wildly inflated to take advantage of those in an already expensive hobby.
And considering they are only executable within PI which is already expensive too. It would be maybe more justified with blurx was a standalone program too. Speculation on my part but i am sure there is more collaboration between rc astro and pixinsight than it appears on the surface...
I have installed the current version under windows 11 and copied a 32bit non-linear image (tiff) into the input folder. after starting the program, the chunks run through and then the program exits. Nothing can be found in the output folder. Even without Cuda, the folder remains empty. Is there a log file for error checking or does anyone have an idea? Incidentally, the test image is processed correctly.
then you have the wrong type of file (32bit can either be unsigned integer or floating point, you need floating point). Be sure you grab the latest version too. I just put it out V4 today.
Very cool! I'm hoping to test this out over the weekend. You need to share your secret on how you've been churning out so many apps/scripts so quickly.
Apologies Frank, but maybe a dumb question but how exactly do I save an on screen PI image as a 32 bit TIFF to then try Cosmic Clarity.....maybe a bit "slow" this morning....thanks
If you have PI just use the repository to get the script interface. It will do all the work for you. Otherwise just make sure it is stretched and go File->Save As
@@setiv2 Hi Frank...OK stuck again,,when I select either of the 32bit choices I get this Pixinsight API Error MetaFileFormat......TIFF Invalid or non-existant directory: /input/Final_1.tiff...Final_1.Tiff is my image on PI screen....No Idea!!!!! what to do next???
Thank you for this !!, I don't have Pix and use Topaz Denoise AI, but Topaz D AI cannot be bought seperately anymore as is bundled in Photo AI, also it has a habit of chewing stars as it was not made for astro work.
Depends. If you are running from PI the script deletes all input and output files. If running stand alone then yes you have to maintain your input and output files
Very excited about this as a non-PixInsight user. Looks like a very useful tool. I've tried it on a couple of images and it's not working for me though. Everything runs, I get to select sharpness levels and I see the processing occurring in the command window, but the output I get every time is a blank, dark grey image. My input images are non-linear 32 bit tiff. Is there anything I can do or any more info I can provide to help diagnose the issue?
@@setiv2 will try the new model. I tried a TIF exported from Siril instead of Affinity and it worked - although it did make the image a lot brighter. Maybe a coloursplace issue?
I dont know if I can or not. I will need to look into that. I tried making xisf support but if you are already using pixinsight using the script wont lose any data so I abandoned a couple days wasted just trying to get that to function.
BXT should only work with linear images. Currently, can Cosmic Clarity only be used for non-linear images? Looking forward to a version for linear images!
It took rc-astro a long time for it to work on linear data. It would always stretch the data behind the scenes, sharpen, and unstretch it. I am not 100% convinced that isnt what it is still doing since he doesnt share his code. In the mean time, if you have PI if you check linear data, it will stretch, sharoen, and unstretch back to a linear state for you
A beginner's question (using Siril for now). At what point in the workflow would this be used? I.e. Before/after denoising, color calibration, etc. Thanks for all you do.
Since we are in early days I would say experiment around. My thought is after color calibration, denoising, and initial stretch. Then sharpen with Cosmic Clarity
Just want to ask a question,,,, I am using your sharpening for the first time. It is taking a super long time. Is that normal ?? its got to be 20 mins now maybe more. Thanks
@@Steve_The_Ignorant_Astronomer of you arent using gpu acceleration and you have a big image then yes. Do you have an nvidia graphics card? You will need CUDA gpu acceleration otherwise it will take a long time on th3 cpu. All these neural net toolsare like that. BXT, NXT, Graxpert Denoise, StarNet, etc
Hello, I downloaded the file for mac, but it’s a zip file. I’m having trouble installing it as it is not a .pkg. I’m not great with computer stuff, so I would appreciate any direction on how to install it on a mac. Cheers!
Sure! There is nothing to install. Just double click the zip file and it will be extracted for you. In that extracted folder there is an inout folder and an output folder along with a file called setiastrocosmicclaritymac . There are also some files that end in .pth. those are the ai models. You can get the latest models from my website too, you can just replace those ones. So what you will want to do is put your image file you want sharpened in the inout folder. Then double click the setiastrocowmiclaritymac file. This will start the program. When you are all done the sharpened file will be in yhe output folder. You can always email me screenshots if you run into issues too. frank@setiastro.com
I'm running on LinuxMint and I have Wine(Windows Emulator) installed on my linux environment. I downloaded the Windows standalone version of Cosmic Clarity. Extracted the files into a folder on the C_Drive of the Wine envorionment and Cosmic Clarity ran just fine simply by executing the .exe file. The test image was processed and the output image file created. It did pop up the initial questions and responded as expected.
Linux Version of Cosmic Clarity Sharpen and Denoise compiled!!! The stand alone is up and running. I still have to fully vet the script interface with PixInsight but that shouldn't stop anyone from using the standalone! Site: www.setiastro.com/cosmic-clarity Drive: drive.google.com/drive/folders/1hB2qeMF7YQH1-Vr81udvNfd_NB6RKBT0?usp=drive_link
@@setiv2 I gave this a try but I get the following message ./SetiAstroCosmicClarity_denoise [PYI-191074:ERROR] Failed to load Python shared library '/home/johnm/CosmicClarity/_internal/libpython3.12.so.1.0': dlopen: /lib/x86_64-linux-gnu/libm.so.6: version `GLIBC_2.38' not found (required by /home/johnm/CosmicClarity/_internal/libpython3.12.so.1.0) The libm.so.6 is in the /lib/x86_64-linux-gnu folder but on Linuxmint version 21.3 GLIBC is only version 2.35. Short of upgrading to Linuxmint 22.0 ( and I'm not sure even that would give me GLIBC 2.38) there doesn't appear to be a easy path to upgrading my GLIBC to 2.38 without causing other issues. If you have any ideas on how to get around this please let me know
@@setiv2 Sorry for the slow response to testing this but I ran into a few issues that took some time to resolve. In my environment i'm using Linux Distro: LinuxMint 21.3 On my first attempt to run SetiAstroCosmicClarity I received the following error message :~/SetiAstro/CosmicClaritySharpenDenoise_LINUX$ ./SetiAstroCosmicClarity [PYI-8442:ERROR] Failed to load Python shared library '/home/johnm/SetiAstro/CosmicClaritySharpenDenoise_LINUX/_internal/libpython3.12.so.1.0': dlopen: /lib/x86_64-linux-gnu/libm.so.6: version `GLIBC_2.38' not found (required by /home/johnm/SetiAstro/CosmicClaritySharpenDenoise_LINUX/_internal/libpython3.12.so.1.0) This error occurs because Linuxmint 21.3 is built with GLIBC_2.35 and this code is not compatible with apps built with GLIBC_2.38. After some internet research I found the best solution was to upgrade to LinuxMint 22.0 which I had planned to do in any case. Once I upgraded my system SetiAstroCosmicClarity got past the GLIBC issue but reported that "libcudnn_engines_precompiled.so.9" was missing which I found in your ~/_internal/nidia/cudnn/lib folder so I copied it to /_internal. SetiAstroCosmicClarity then reported another libcudnn... file missing so I copied all the libcudnn files in ~/_internal/nidia/cudnn/lib to /_internal that where not already in _internal folder. SetiAstroCosmicClarity and SetiAstroCosmicClarity_denoise then ran to completion and used the GPU acceleration which greatly improved the performance. Hope this helps as I believe a number of us Astro photograghers would like to switch over to running on some Linux distro. With this I can do all my Astro photographic processing using Siril, SetiAstroCosmicClarity and GIMP to process my images. Thanks John M.
I have been watching your videos and downloaded the scripts , thx for all the work you do. I have a question on Pix , I have not been processing for about 6 months and been having a bit of trouble and was wondering if you can help me out with a problem i am having. I do DBE and sharpening and it looks good but when I reset the stretch and go to a black image and stretch again my image looks all wonky but if I use the back arrow to go back 2 steps and unstretch and do an stf again it looks fine... Do you have any idea what I am doing wrong... Thx for all the scripts again
You are doing nothing wrong at all. You should checkout my video on STF ruclips.net/video/lo1SssohJgc/видео.htmlsi=T-gy-ZePVDRrDGCh It is a big reason why I made my Statistical Stretch. You need to check out those videos too :)
@@setiv2 I never had this problem in the past It's almost like I'm double stretching. I use noise exterminator twice and it looks nice and smooth and then he gets wonky It's just so freaking annoying
@@setiv2 okay I did that statistical stretch and it worked good but But I'm not sure what I was doing wrong the old way which I used to do it I must be doing something wrong a step wrong or something and when I'm trying to do a histogram background it gets wonky now So something I'm not doing right but at least that statistical stretch works really well
stars within nebula and galaxies are good data if you want to contribute. so that way when i train stellar and non-stellar it can determine one from the other. i'll take any data anyone wants to contribute :)
@@setiv2 Of course! Maybe if you set up an open Google drive folder (or whatever solution works for you), people can add their work there. 👍 I'd happily add my modest amount of OSC data (don't have a mono camera yet).
Getting a time out error running your script in PixInsight non cuda setup, but it still sharpens the image when reloaded from the output folder back in to Pix.
oh that is something I didn't think. non cuda probably takes more than 2 minutes to process. the timeout was set to 2 minutes. it should still be accessible like you found from the output folder when it is completed.
Can't get your mac version to work how does it start up? something is missing this is the error message "“Python.framework” is damaged and can’t be opened. You should move it to the Bin."
i have not been able to recreate that error. you extracted everything in the folder and didn't move any of the file out of the extracted folder did you? it should all be contained in the SetiAstroCosmicClarity folder still after extraction
@@setiv2 yes but how exactly does it startup? ie how do I make it work as clicking on the terminal app is the only way of starting it? Then I get that error
the info.plist shows this file link "file://localhost/System/Library/DTDs/PropertyList.dtd"> I think it looks for a file that doesnt exist, this is in the python resources I also think it's looking for a file in your folder on your device? I'm not sure I'm not a programmer
Not sure what you mean. You have to download the cosmicclarity program first, extract it. Then if you use pixinsight get the repository for the script to interface with the program
Is there something that gradient removal isnt getting rid of? Smooth flats usually arent hard, the difficult thing you truly need flats for are dust donuts and other aberrations like that
BlurX has so many parameters. Near as I can tell there is at least a correction ai model, a stellar sharpening ai model, a halo sharpening model, a non stellar sharpening model, a psf model, and each of those probably have various sub modelling for the different strength/psf levels. It was also trained across a whole gambit of aberrations from both professional and amateur equipement. Like I said, the RollsRoyce program vs my gold cart :)
Problems running the script. The console opens, but no choice window follows. Then I see a very fast error message in the cosole and it closes down. Reinstalled, same problem.
I also just uploaded an updated zip folder. It wasn't catching if you don't have CUDA and erroring out. It should revert to CPU now if it doesn't find CUDA acceleration
Great job. Congratulations. I can't run the script. It gives me this error. RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU. [PYI-22976:ERROR] Failed to execute script 'SetiAstroCosmicClarity' due to unhandled exception! I use an AMD graphics card, it could be because it does not work with CUDA. I eagerly await your response. Thank you
Be sure you got the latest version: drive.google.com/file/d/1WQvQvrrLFgzbZwe-exeHjECmawHKYGyS/view?usp=drive_link it should default back to CPU if yuou don't have CUDA.
If this is truly an AI, it can't be trusted. It may give you wonderful pictures, but you will never know if it has added or removed details. So, don't use it for research, just for pretty pictures.
To be fair this is the ONLY tool I have made so far that doesn't work on Linux. Even BlindSolve2000 works on linux now. Also this is a hobby afterall and I am not in software development. To learn a 3rd operating system and get it installed on a personal computer is just going to take some time.
@@KurtFB-r6x 😮💨 Always those comments from people who believe they know but don't know that they.. believe only. I bought a computer and install Linux to connect to internet in 1998 because it was already efficient and reliable (unlike windows) and since, because I don't like being chained to a brand (Apple). I've never had a virus, a blue screen or a crash since, and never had to reinstall a distribution due to a bug. I control my computer and only I. Can you imagine that :))
Great to see this! I'm working in a MacOS and no problem with denoise tool, but when I try running sharpening tool I have a crash and this message in the Terminal " Error reading image /Users/ramiro/Desktop/Seestar/CosmicClarity/input/.DS_Store: cannot identify image file '/Users/ramiro/Desktop/Seestar/CosmicClarity/input/.DS_Store' Traceback (most recent call last): File "setiastrocosmicclaritymac.py", line 857, in File "setiastrocosmicclaritymac.py", line 760, in process_images ValueError: not enough values to unpack (expected 4, got 3) [PYI-4780:ERROR] Failed to execute script 'setiastrocosmicclaritymac' due to unhandled exception! Any help? Doing inside PI I dont get the message but fails adter a long time. Thanks in advance!
@@setiv2 Thanks, Now it's working! .DS_Store seems to be an automatic file created on MacOS every time you open a folder containing visual information.
Newest modelling loaded. trained on a deeper neural net and added more data. I need to find a better way to telling the broader audience when a new version is available.
Windows Version: drive.google.com/file/d/1WQvQvrrLFgzbZwe-exeHjECmawHKYGyS/view?usp=drive_link
MacOS Version: drive.google.com/file/d/1Tze3E6b_si8ZjUXkPMj8TICqrP88ulbJ/view?usp=drive_link
The astronomy/astro community owes you a great big thank you! And coffee donations.... for sure.
Another great effort on your part Frank. I know you refer to this as the "golf cart" compared to the "gold standard", but I think this can only promote constructive competition amongst developers that in the long run will benefit everyone. Thanks.
Your work is so amazing it’s scary! I can only imagine this new tool as it continues to learn.
This is awesome! I'm so glad people are making free tools like this! Great work! :)
You should shoot me an email, I know you are working a sharpening tool as well. I'm more than happy to collaborate and share my code on this! I think you are much more knowledgeable in this than I am. frank@setiastro.com
You are a machine Frank ! Thanks so much for yet another awesome tool.
Wow, you are churning these superb scripts out…👏🏻👏🏻👍🏻
Great tool. Many thanks! 😍
Seti-Sharpen... Missed naming opportunity. 😅. Awesome work as always.
Still a beta, we can change it :) I did have it Seti Astro Sharpen, but that is too close to Deep Sky Detail's sharpen tool he is working on AstroSharp
Very good tool! Thanks Fran. A question, for linux?
Thank you so much! Imma try this asap!
You're amazing, thank you. While the RC astro suite is outstanding. The price is wildly inflated to take advantage of those in an already expensive hobby.
And considering they are only executable within PI which is already expensive too. It would be maybe more justified with blurx was a standalone program too. Speculation on my part but i am sure there is more collaboration between rc astro and pixinsight than it appears on the surface...
Congratulations.. and Thanks
Thanks for a great sharpening tool. However, I am curious about the reason for a 2.5G Windows file vs 400MB for the Mac file.
Mac has native gpu acceleration. Windows needs the CUDA and torch support for gpu acceleration that gets bundled in the distribution.
I have installed the current version under windows 11 and copied a 32bit non-linear image (tiff) into the input folder. after starting the program, the chunks run through and then the program exits. Nothing can be found in the output folder. Even without Cuda, the folder remains empty. Is there a log file for error checking or does anyone have an idea?
Incidentally, the test image is processed correctly.
then you have the wrong type of file (32bit can either be unsigned integer or floating point, you need floating point). Be sure you grab the latest version too. I just put it out V4 today.
Very cool! I'm hoping to test this out over the weekend.
You need to share your secret on how you've been churning out so many apps/scripts so quickly.
Apologies Frank, but maybe a dumb question but how exactly do I save an on screen PI image as a 32 bit TIFF to then try Cosmic Clarity.....maybe a bit "slow" this morning....thanks
If you have PI just use the repository to get the script interface. It will do all the work for you. Otherwise just make sure it is stretched and go File->Save As
@@setiv2 Hi Frank...OK stuck again,,when I select either of the 32bit choices I get this Pixinsight API Error MetaFileFormat......TIFF Invalid or non-existant directory: /input/Final_1.tiff...Final_1.Tiff is my image on PI screen....No Idea!!!!! what to do next???
@@lenjones2268 you need to use the wrench icon to select the folder your cosmic clarity is in
Great tool. Keep growing! Channel subscription done!
Very good! Thanks man!
Thank you for this !!, I don't have Pix and use Topaz Denoise AI, but Topaz D AI cannot be bought seperately anymore as is bundled in Photo AI, also it has a habit of chewing stars as it was not made for astro work.
Do you need to maintain and clean the output directory manually?
Depends. If you are running from PI the script deletes all input and output files. If running stand alone then yes you have to maintain your input and output files
Very excited about this as a non-PixInsight user. Looks like a very useful tool. I've tried it on a couple of images and it's not working for me though. Everything runs, I get to select sharpness levels and I see the processing occurring in the command window, but the output I get every time is a blank, dark grey image. My input images are non-linear 32 bit tiff. Is there anything I can do or any more info I can provide to help diagnose the issue?
Can you share you input file with me? You can email me frank@setiastro.com
Update: PNG format works for me, but TIF does not. I'm creating both formats by exporting from Affinity Photo. Might be something to do with Affinity?
@@captainfruitbatify potentially. If png is working that is a good start. Version 2 just released as well with a better AI model
@@setiv2 will try the new model. I tried a TIF exported from Siril instead of Affinity and it worked - although it did make the image a lot brighter. Maybe a coloursplace issue?
@@setiv2 Done
can you add FITS support. pulling out files to TIFF for external processing is always a pain as loose context information for the image.
I dont know if I can or not. I will need to look into that. I tried making xisf support but if you are already using pixinsight using the script wont lose any data so I abandoned a couple days wasted just trying to get that to function.
BXT should only work with linear images. Currently, can Cosmic Clarity only be used for non-linear images? Looking forward to a version for linear images!
It took rc-astro a long time for it to work on linear data. It would always stretch the data behind the scenes, sharpen, and unstretch it. I am not 100% convinced that isnt what it is still doing since he doesnt share his code. In the mean time, if you have PI if you check linear data, it will stretch, sharoen, and unstretch back to a linear state for you
A beginner's question (using Siril for now). At what point in the workflow would this be used? I.e. Before/after denoising, color calibration, etc. Thanks for all you do.
Since we are in early days I would say experiment around. My thought is after color calibration, denoising, and initial stretch. Then sharpen with Cosmic Clarity
I’ve been waiting for a free tool like this. Thank you!
Many thanks, great job 👏👏👏👏👏
I don't have PI. It just confuses me. Thank you for making things for us non PI folks who still want to make pretty pictures :)
Just want to ask a question,,,, I am using your sharpening for the first time. It is taking a super long time. Is that normal ?? its got to be 20 mins now maybe more. Thanks
@@Steve_The_Ignorant_Astronomer of you arent using gpu acceleration and you have a big image then yes. Do you have an nvidia graphics card? You will need CUDA gpu acceleration otherwise it will take a long time on th3 cpu. All these neural net toolsare like that. BXT, NXT, Graxpert Denoise, StarNet, etc
I´ve noticed, that my antivirus Avast is blocking the program and no image appears in output folder.
You will need to whitelist the program in your antivirus
@@setiv2 I did it.
Cant wait to try it ty
looking forward to the Mac version!
@@M31glow mac version ready for standalone testing. I will work to get it incorporated into the PI script later. I pinned a comment with the link
Hello, I downloaded the file for mac, but it’s a zip file. I’m having trouble installing it as it is not a .pkg. I’m not great with computer stuff, so I would appreciate any direction on how to install it on a mac. Cheers!
Sure! There is nothing to install. Just double click the zip file and it will be extracted for you. In that extracted folder there is an inout folder and an output folder along with a file called setiastrocosmicclaritymac . There are also some files that end in .pth. those are the ai models. You can get the latest models from my website too, you can just replace those ones. So what you will want to do is put your image file you want sharpened in the inout folder. Then double click the setiastrocowmiclaritymac file. This will start the program. When you are all done the sharpened file will be in yhe output folder. You can always email me screenshots if you run into issues too. frank@setiastro.com
I'm running on LinuxMint and I have Wine(Windows Emulator) installed on my linux environment. I downloaded the Windows standalone version of Cosmic Clarity. Extracted the files into a folder on the C_Drive of the Wine envorionment and Cosmic Clarity ran just fine simply by executing the .exe file. The test image was processed and the output image file created. It did pop up the initial questions and responded as expected.
Oh that is excellent! That is my next step is getting a linux virtual environment so I can port it over to linux
Linux Version of Cosmic Clarity Sharpen and Denoise compiled!!! The stand alone is up and running. I still have to fully vet the script interface with PixInsight but that shouldn't stop anyone from using the standalone!
Site: www.setiastro.com/cosmic-clarity
Drive: drive.google.com/drive/folders/1hB2qeMF7YQH1-Vr81udvNfd_NB6RKBT0?usp=drive_link
@@setiv2 I gave this a try but I get the following message
./SetiAstroCosmicClarity_denoise
[PYI-191074:ERROR] Failed to load Python shared library '/home/johnm/CosmicClarity/_internal/libpython3.12.so.1.0': dlopen: /lib/x86_64-linux-gnu/libm.so.6: version `GLIBC_2.38' not found (required by /home/johnm/CosmicClarity/_internal/libpython3.12.so.1.0)
The libm.so.6 is in the /lib/x86_64-linux-gnu folder but on Linuxmint version 21.3 GLIBC is only version 2.35.
Short of upgrading to Linuxmint 22.0 ( and I'm not sure even that would give me GLIBC 2.38) there doesn't appear to be a easy path to upgrading my GLIBC to 2.38 without causing other issues. If you have any ideas on how to get around this please let me know
@@setiv2
Sorry for the slow response to testing this but I ran into a few issues that took some time to resolve.
In my environment i'm using Linux Distro: LinuxMint 21.3
On my first attempt to run SetiAstroCosmicClarity I received the following error message
:~/SetiAstro/CosmicClaritySharpenDenoise_LINUX$ ./SetiAstroCosmicClarity
[PYI-8442:ERROR] Failed to load Python shared library '/home/johnm/SetiAstro/CosmicClaritySharpenDenoise_LINUX/_internal/libpython3.12.so.1.0': dlopen: /lib/x86_64-linux-gnu/libm.so.6: version `GLIBC_2.38' not found (required by /home/johnm/SetiAstro/CosmicClaritySharpenDenoise_LINUX/_internal/libpython3.12.so.1.0)
This error occurs because Linuxmint 21.3 is built with GLIBC_2.35 and this code is not compatible with apps built with GLIBC_2.38.
After some internet research I found the best solution was to upgrade to LinuxMint 22.0 which I had planned to do in any case.
Once I upgraded my system SetiAstroCosmicClarity got past the GLIBC issue but reported that "libcudnn_engines_precompiled.so.9" was missing which I found in your ~/_internal/nidia/cudnn/lib folder so I copied it to /_internal. SetiAstroCosmicClarity then reported another libcudnn... file missing so I copied all the libcudnn files in ~/_internal/nidia/cudnn/lib to /_internal that where not already in _internal folder.
SetiAstroCosmicClarity and SetiAstroCosmicClarity_denoise then ran to completion and used the GPU acceleration which greatly improved the performance.
Hope this helps as I believe a number of us Astro photograghers would like to switch over to running on some Linux distro. With this I can do all my Astro photographic processing using Siril, SetiAstroCosmicClarity and GIMP to process my images.
Thanks John M.
I have been watching your videos and downloaded the scripts , thx for all the work you do. I have a question on Pix , I have not been processing for about 6 months and been having
a bit of trouble and was wondering if you can help me out with a problem i am having. I do DBE and sharpening and it looks good but when I reset the stretch and go to a black image and
stretch again my image looks all wonky but if I use the back arrow to go back 2 steps and unstretch and do an stf again it looks fine... Do you have any idea what I am doing wrong...
Thx for all the scripts again
You are doing nothing wrong at all. You should checkout my video on STF ruclips.net/video/lo1SssohJgc/видео.htmlsi=T-gy-ZePVDRrDGCh
It is a big reason why I made my Statistical Stretch. You need to check out those videos too :)
@@setiv2 I never had this problem in the past It's almost like I'm double stretching. I use noise exterminator twice and it looks nice and smooth and then he gets wonky It's just so freaking annoying
@@setiv2 I will check out statistical stretch and see how I do
@@Steve_The_Ignorant_Astronomer that is how stf functions.
@@setiv2 okay I did that statistical stretch and it worked good but But I'm not sure what I was doing wrong the old way which I used to do it I must be doing something wrong a step wrong or something and when I'm trying to do a histogram background it gets wonky now So something I'm not doing right but at least that statistical stretch works really well
Love it! Is there anything we can do to help here? Maybe contribute more data?
stars within nebula and galaxies are good data if you want to contribute. so that way when i train stellar and non-stellar it can determine one from the other. i'll take any data anyone wants to contribute :)
@@setiv2 Of course! Maybe if you set up an open Google drive folder (or whatever solution works for you), people can add their work there. 👍 I'd happily add my modest amount of OSC data (don't have a mono camera yet).
Awesome, thank you!
Getting a time out error running your script in PixInsight non cuda setup, but it still sharpens the image when reloaded from the output folder back in to Pix.
oh that is something I didn't think. non cuda probably takes more than 2 minutes to process. the timeout was set to 2 minutes. it should still be accessible like you found from the output folder when it is completed.
@@setiv2 Yes that's correct and thank you for your fantastic work
Can't get your mac version to work how does it start up? something is missing this is the error message "“Python.framework” is damaged and can’t be opened. You should move it to the Bin."
Ok let me look at that
i have not been able to recreate that error. you extracted everything in the folder and didn't move any of the file out of the extracted folder did you? it should all be contained in the SetiAstroCosmicClarity folder still after extraction
@@setiv2 yes but how exactly does it startup? ie how do I make it work as clicking on the terminal app is the only way of starting it? Then I get that error
Same error: "“Python.framework” is damaged and can’t be opened.". Did you create the executable for arm64 or just for x86?
the info.plist shows this file link "file://localhost/System/Library/DTDs/PropertyList.dtd"> I think it looks for a file that doesnt exist, this is in the python resources I also think it's looking for a file in your folder on your device? I'm not sure I'm not a programmer
In which folder is it installed?
Not sure what you mean. You have to download the cosmicclarity program first, extract it. Then if you use pixinsight get the repository for the script to interface with the program
@@setiv2 You must select a folder with the wrench.
@@setiv2 Sorry, I thought the script version for pixinsight was complete with everything, not that I had to download the ZIP file too.
@@PaoloBanci no you need to download the zip too. Similar to like you need to download graxpert and the script in PI just interfaces with it
Can you creat a script for synthetics flats?
Is there something that gradient removal isnt getting rid of? Smooth flats usually arent hard, the difficult thing you truly need flats for are dust donuts and other aberrations like that
You should create a new set of training files by using blurx to create the training data !!
BlurX has so many parameters. Near as I can tell there is at least a correction ai model, a stellar sharpening ai model, a halo sharpening model, a non stellar sharpening model, a psf model, and each of those probably have various sub modelling for the different strength/psf levels. It was also trained across a whole gambit of aberrations from both professional and amateur equipement. Like I said, the RollsRoyce program vs my gold cart :)
Problems running the script. The console opens, but no choice window follows. Then I see a very fast error message in the cosole and it closes down. Reinstalled, same problem.
Please try running it from the cmd instead of double-clicking the exe file. This way, the window should remain open.
I also just uploaded an updated zip folder. It wasn't catching if you don't have CUDA and erroring out. It should revert to CPU now if it doesn't find CUDA acceleration
Great job. Congratulations. I can't run the script. It gives me this error. RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.
[PYI-22976:ERROR] Failed to execute script 'SetiAstroCosmicClarity' due to unhandled exception!
I use an AMD graphics card, it could be because it does not work with CUDA. I eagerly await your response. Thank you
Be sure you got the latest version: drive.google.com/file/d/1WQvQvrrLFgzbZwe-exeHjECmawHKYGyS/view?usp=drive_link it should default back to CPU if yuou don't have CUDA.
Do you sleep? Awesome.
Very nice! Finally some real alternative to the stupidly overpriced blurx plugin...
Unlucky max ;)
In pixinsight it doesn’t works.
What doesnt work? Did you download the main program first and use the wrench icon to tell the script where the cosmic clarity program is?
If this is truly an AI, it can't be trusted. It may give you wonderful pictures, but you will never know if it has added or removed details. So, don't use it for research, just for pretty pictures.
I am assuming you dont use blurx, noisex, deepsnr, or any others then, they all run convolution neural nets on tensorflow
You are doing free tools. Why don't you develop on and make Linux version ;(
Beggars can’t be choosers
To be fair this is the ONLY tool I have made so far that doesn't work on Linux. Even BlindSolve2000 works on linux now. Also this is a hobby afterall and I am not in software development. To learn a 3rd operating system and get it installed on a personal computer is just going to take some time.
@@KurtFB-r6x 😮💨 Always those comments from people who believe they know but don't know that they.. believe only.
I bought a computer and install Linux to connect to internet in 1998 because it was already efficient and reliable (unlike windows) and since, because I don't like being chained to a brand (Apple).
I've never had a virus, a blue screen or a crash since, and never had to reinstall a distribution due to a bug. I control my computer and only I.
Can you imagine that :))
@@setiv2 I understand. Anyway thanks for your great work 🙂
Great to see this! I'm working in a MacOS and no problem with denoise tool, but when I try running sharpening tool I have a crash and this message in the Terminal " Error reading image /Users/ramiro/Desktop/Seestar/CosmicClarity/input/.DS_Store: cannot identify image file '/Users/ramiro/Desktop/Seestar/CosmicClarity/input/.DS_Store'
Traceback (most recent call last):
File "setiastrocosmicclaritymac.py", line 857, in
File "setiastrocosmicclaritymac.py", line 760, in process_images
ValueError: not enough values to unpack (expected 4, got 3)
[PYI-4780:ERROR] Failed to execute script 'setiastrocosmicclaritymac' due to unhandled exception!
Any help? Doing inside PI I dont get the message but fails adter a long time.
Thanks in advance!
You need tp delete that hidden file .DS_Store in your inout folder. It thinks that is an image
@@setiv2 Thanks, Now it's working! .DS_Store seems to be an automatic file created on MacOS every time you open a folder containing visual information.
@@ramirohuete ridoculous!