143 replies to this topic

[schmelly]

Dear community,

 

I am one of the developers of GraXpert, a free and open source (BSD-3-licensed) standalone tool that makes removal of gradients in deep-sky photos easy and free, and provides - as we believe - high quality results. We just released GraXpert (RC1) for the first time earlier this week. We provide both the source code as well as binary releases for Linux (tested on Ubuntu 20.04), Windows (>10), and MacOS (>10.15) on github.

 

The project: https://github.com/Steffenhir/GraXpert

Releases: https://github.com/S...aXpert/releases

The license: https://github.com/S...ob/main/LICENSE

 

The tool is from hobbyists for hobbyists and we do not make any money out of it.

 

--------------------------------------------------

A note to the moderatos of this forum: with this introduction, I hope I do meet the requirements on mentioning open source software here, following your "Software Developers Please Read" topic: https://www.cloudyni...rs-please-read/

If you think my topic is ill-placed I would be more than happy to get the vendor status flag (I assume it is free for open source software?) and that you move my topic to the vendor announcement forum.

--------------------------------------------------

 

Maybe a bit of background information on the tool: the idea for GraXpert was born by Frank Sackenheim in context of the Dark Matters Discord community about three months ago. The idea was to create a (very!) affordable tool for gradient removal, especially targeting beginners. Steffen, Christian, and me picked up the idea and we were able to produce a first release candidate for which we would love to hear your feedback and see your results!

 

Frank is a german astro photographer, a member of the Dark Matters community, and he was kind enough to setup a website, https://www.graxpert.com, that provides both a video introduction (in English and German), a written user manual, and also the possibility to download GraXpert from there.

 

Do you want to get in contact with us? Please feel free to join the Dark Matters community where we have dedicated channels for GraXpert. Maybe there is a lot of German going on right now. However, generally at Dark Matters we use both English and German so do not hesitate to start a discussion

https://discord.gg/9gSq8QCH8g

 

Do you want to open bug reports, feature requests, or even want to contribute? Please have a look at our github project linked above.

 

I would like to thank my fellow developers for their huge time investments during the last two months making this possible in the first place, the Dark Matters community for great contributions (the manual, the logo, the beta testing, the feedback, ...) and Frank for bringing us together in the first place

 

 

We hope you like it and we wish you a lot of fun getting rid of those shabby gradients (for free )!

David

Edited by schmelly, 27 April 2022 - 06:58 AM.

[OldManSky]

I will have a look at it.

Whether I use it or not, I want to thank the developers for the time and effort put into the tool.  It's people like you who help keep our community growing and our hobby getting better.

Thanks for the work and for making it free/open source!  I've been a programmer/manager for 40 years, so I know how much work goes into a tool like this!

[flamidey]

This is a very good initiative. Gradient removal is really not beginner friendly and this looks very promising.

 

Thank you!

[bobharmony]

I browsed through the manual and like the look of things.  I will be kicking the tires on this as gradients tend to be the worst issue I deal with having lots of local light domes around my imaging spot.

 

Thanks for taking the time to develop and share what looks to be a promising tool.

 

Bob

[schmelly]

Dear community,

 

a friendly community member pm'ed me and told me that Bitdefender antivirus is flagging Graxpert as suspicious and I would like to be open and comment on her finding:

 

I am pretty sure her finding is a false positive and has something to do with the way we bundle the GraXpert application. Let me explain: Technically, we develop GraXpert in the programming language Python and bundle it via a tool called PyInstaller as an executable file. At application startup, Windows unpacks the executable in a temporary directory and the bundled Python runtime starts the actual application from there. This unpacking+execution pattern is similar to how some malwares behave and I suspect some scanners find this suspicious.

 

From a development standpoint, we perform two measures to make the application as safe for the users as we can:

• The binary distribution of GraXpert is "compiled" on github infrastructure, i.e., in virtual machines hosted by github, exclusively. That is, in particular we do not publish binaries that have been created on our local developer machines (which might be infected - in the end, we cannot be sure of the opposite).
• Before we announce a release to the public we submit it for malware analysis to Microsoft. If we wouldn't do that the Microsoft Defender SmartScreen Application would kick in and would flag the file as unknown / suspicious during application startup. I attached the Microsoft analysis results of our RC1 release from 2022-04-22 for your reference.

As a security researcher in software engineering I generally recommend that you do not trust any software provided on the internet without further checks. I can recommend virustotal.com for this. It is an online malware scanner which accepts files up to 500mb and executes a whole bunch of malware scanning engines on your submitted file for free. If you see only few positive (red) results, for me this is generally a good approximation if a suspicious file is a false positive or not.

 

If you do not trust our binary distribution that is ok. In that case you still have the opportunity to use Graxpert based on the source distribution. We are happy to help on how to to this (e.g., in the Dark Matters Discord (cf. my first post) or here).

 

Kind regards

David

 

Attached Thumbnails

• 

[alphatripleplus]

Everyone please bear with us as we review this topic. Thanks for your patience....

 

Edit: Back to our regular programming .... please continue. 

[arbit]

I tried it on an image with really terrible and complex gradients (various reasons).

 

The results were quite nice. The default RBF was okay. I noticed that the Splines algorithm gave a result similar to DBE (at least similar to DBE order 2). Kriging was the slowest but also the best, as advertised.

 

The strong points (for me) are

1. Easy UI, with a very useful help screen always available

2. Focus on just 4 key parameters (which are the ones I mostly change in DBE anyway)

3. Choice of algorithms 

 

Suggestions

1. Crop facility - I know this can be done elsewhere, but crop/gradient removal is usually the first step so it would be convenient

2. if possible, support xisf - again, TIFF and FITS are universal, but this is useful for PI users.

 

Thanks for the software. The UX is really nice.

 

Edit: Bitdefender flags the download from the website but not from Github! Also, VirusTotal passes it.

Edited by arbit, 01 May 2022 - 09:00 AM.

[frasax]

I tried it on an image with really terrible and complex gradients (various reasons).

 

The results were quite nice. The default RBF was okay. I noticed that the Splines algorithm gave a result similar to DBE (at least similar to DBE order 2). Kriging was the slowest but also the best, as advertised.

 

The strong points (for me) are

1. Easy UI, with a very useful help screen always available

2. Focus on just 4 key parameters (which are the ones I mostly change in DBE anyway)

3. Choice of algorithms 

 

Suggestions

1. Crop facility - I know this can be done elsewhere, but crop/gradient removal is usually the first step so it would be convenient

2. if possible, support xisf - again, TIFF and FITS are universal, but this is useful for PI users.

 

Thanks for the software. The UX is really nice.

 

Edit: Bitdefender flags the download from the website but not from Github! Also, VirusTotal passes it.

We have discussed the implementation of xisf a couple of times. While i thought it is not a big deal, it turned out it is a bit more than that:-) 
What would be the actual benefits in a PixInsight Workflow, except for having the naming consistent?

CS Frank 

[arbit]

We have discussed the implementation of xisf a couple of times. While i thought it is not a big deal, it turned out it is a bit more than that:-)
What would be the actual benefits in a PixInsight Workflow, except for having the naming consistent?

CS Frank

Hi, both crop and xisf are conveniences really.

In a PI workflow the raw stack output after integration will be xisf.

So to use Graxpert, it will have to be cropped as xisf, then converted to fits, opened and processed in GE, saved as fits, then opened and saved as xisf in PI.

So direct support in GE for crop and xisf makes it simpler. Open directly in GE, crop, process, save and open directly in PI without any conversions.

Of course, one can always ask why not just use fits even in PI, but that's a different discussion :-)

Edit: I should clarify that IMO these are relatively minor inconveniences, and nice to have. if time is limited, and these are time consuming changes, then as a user higher priority would be to including more algorithms, performance etc. Hope that helps.

Sent from my SM-N960F using Tapatalk

Edited by arbit, 01 May 2022 - 08:38 PM.

[frasax]

Hi, both crop and xisf are conveniences really.

In a PI workflow the raw stack output after integration will be xisf.

So to use Graxpert, it will have to be cropped as xisf, then converted to fits, opened and processed in GE, saved as fits, then opened and saved as xisf in PI.

So direct support in GE for crop and xisf makes it simpler. Open directly in GE, crop, process, save and open directly in PI without any conversions.

Of course, one can always ask why not just use fits even in PI, but that's a different discussion :-)

Edit: I should clarify that IMO these are relatively minor inconveniences, and nice to have. if time is limited, and these are time consuming changes, then as a user higher priority would be to including more algorithms, performance etc. Hope that helps.

Sent from my SM-N960F using Tapatalk

Hi,

 

i see. Yes, agree with everything here. We will keep that in mind. The crop shouldn´t be a big issue here, but it seems like including xisf needs to have further investigation and is not top priority yet. 

 

Thanks for your input, its highly appreciated.

 

CS Frank

[ranjin]

Very convenient software, I have used it and I feel very good. Suitable for beginners like me. Thank you!

[spames]

Thank you. Works will for me.

[galacticinsomnia]

Looking good.  This really reminds me a lot of the background extraction module in Siril.  Any relation ? 

Good work and thank you for your contributions !

Clear Skies !!

[schmelly]

GraXpert 'Tethys' RC2 (v0.1.3)

 

Summary

 

We are happy to announce the second release candidate for GraXpert 'Thetys' which includes a couple of improvements compared to RC1 (cf. release notes below). A note to our MacOS users: in order to start GraXpert from your applications you should download the .dmg file, copy the contained application to your applications and start it from there. For the first application startup you probably have to allow the execution after your first double click: Go to System Preferences -> Security and Privacy and you should see an entry where you can whitelist GraXpert. The first startup might take a while. Subsequent startups should be considerably faster.

 

Thanks a lot for your comprehensive and positive feedback so far!

 

You can download the release here: https://github.com/S...ases/tag/v0.1.3
(we will update the download links at https://www.graxpert.com/ shortly)

 

Supported Systems

 

We have built and tested GraXpert on:

• Ubuntu 20.04. (i.e., Debian 11 should work as well)
• Windows 10
• MacOS 10.15 (i.e., any MacOS release afterward should work)

Improvements since RC1

• The grid selection now chooses the darkest quadrant around a grid point
• Added options to adjust the size and color of sample points
• Available key bindings are now documented as part of the help panel
• Reduced perceived noise in stretched previews
• Added a panel 'advanced' to adjust RBF and spline settings, the display language, and other settings
• Improved language detection for German language variants like 'de_CH'
• .fts files are now recognized as fits
• We know read and interpret the 'ROWORDER' field of .fits files to avoid their incorrect, flipped display
• Improved format detection for capitalized file endings (e.g., .TIFF)
• Scrollbars are now scrollable via the mouse wheel
• Incorporated a first, self-compiled PyInstaller bootloader. This should at least reduce the false-positive flagging of GraXpert by some malware detection engines on Windows
• GraXpert is now available as .zip bundles for both Windows and Linux which should improve startup time compared to the self-extracting binary distributions
• Improved the startup time on MacOS

Fixed Bugs

• Fixed a bug where GraXpert refused to start in case it encountered an invalid preferences file
• Fixed a bug where GraXpert refused to start during iteration of available monitors and their dpi

Known Issues

• (bug) GraXpert won't start on some Linux systems with multiple monitors

Full Changelog: https://github.com/S...v0.1.2...v0.1.3

 

 

This really reminds me a lot of the background extraction module in Siril.  Any relation?

In addition to spline-based interpolation that Siril includes today we provide two additional interpolation methods (RBF and Kriging) that should provide better results in case of complex gradients.

In the meantime we are in contact with the Siril developers and together we are working on an integration of our algorithms into Siril - stay tuned

 

CS

David

[galacticinsomnia]

GraXpert 'Tethys' RC2 (v0.1.3)

 

Summary

 

We are happy to announce the second release candidate for GraXpert 'Thetys' which includes a couple of improvements compared to RC1 (cf. release notes below). A note to our MacOS users: in order to start GraXpert from your applications you should download the .dmg file, copy the contained application to your applications and start it from there. For the first application startup you probably have to allow the execution after your first double click: Go to System Preferences -> Security and Privacy and you should see an entry where you can whitelist GraXpert. The first startup might take a while. Subsequent startups should be considerably faster.

 

Thanks a lot for your comprehensive and positive feedback so far!

 

You can download the release here: https://github.com/S...ases/tag/v0.1.3
(we will update the download links at https://www.graxpert.com/ shortly)

 

Supported Systems

 

We have built and tested GraXpert on:

• Ubuntu 20.04. (i.e., Debian 11 should work as well)
• Windows 10
• MacOS 10.15 (i.e., any MacOS release afterward should work)

Improvements since RC1

• The grid selection now chooses the darkest quadrant around a grid point
• Added options to adjust the size and color of sample points
• Available key bindings are now documented as part of the help panel
• Reduced perceived noise in stretched previews
• Added a panel 'advanced' to adjust RBF and spline settings, the display language, and other settings
• Improved language detection for German language variants like 'de_CH'
• .fts files are now recognized as fits
• We know read and interpret the 'ROWORDER' field of .fits files to avoid their incorrect, flipped display
• Improved format detection for capitalized file endings (e.g., .TIFF)
• Scrollbars are now scrollable via the mouse wheel
• Incorporated a first, self-compiled PyInstaller bootloader. This should at least reduce the false-positive flagging of GraXpert by some malware detection engines on Windows
• GraXpert is now available as .zip bundles for both Windows and Linux which should improve startup time compared to the self-extracting binary distributions
• Improved the startup time on MacOS

Fixed Bugs

• Fixed a bug where GraXpert refused to start in case it encountered an invalid preferences file
• Fixed a bug where GraXpert refused to start during iteration of available monitors and their dpi

Known Issues

• (bug) GraXpert won't start on some Linux systems with multiple monitors

Full Changelog: https://github.com/S...v0.1.2...v0.1.3

 

 

In addition to spline-based interpolation that Siril includes today we provide two additional interpolation methods (RBF and Kriging) that should provide better results in case of complex gradients.

In the meantime we are in contact with the Siril developers and together we are working on an integration of our algorithms into Siril - stay tuned

 

CS

David

Thank you, and that would be very exciting.   !!

Clear Skies !!

[dciobota]

A big hand to David and the team for developing this very useful app.

 

Some background.  I used to use gradient xterminator with mixed success many years ago.  "Graduated" to PixInsight and got somewhat better results, but the clunky interface and arcane workflow of that app put me off.  Discovered AstroPixelProcessor and love it, and the gradient removal tool was more intuitive to use and worked very well most of the time.

 

However, I ended up taking some milky way frames in a light pollution scenario that created very complex and uneven gradients.  It has been a nightmare to reduce them.  APP did a somewhat ok job.  Then I discovered this thread and downloaded the win64 latest and tried it.  RBF and Kriging didn't do a great job (maybe it's just me using it the first time) but splines actually worked better than the best I could get with APP.

 

Thank you again David and the team, you have rescued my pic!  I don't know if it's appropriate to post a comparison here, so I won't unless requested.

 

Keep up the great work.

[flamidey]

Sadly the new version does not start on my Mac M1 anymore. It has been properly authorized with ctrl+open but the double click on the .app does nothing. 

I can launch it by going inside the package and launching GraXpert from there.

[Delco714]

So take my tiff after stacking into graX? Then save the tiff over and pull it into pixinsight? Thanks!

GraXpert 'Tethys' RC2 (v0.1.3)

Summary

We are happy to announce the second release candidate for GraXpert 'Thetys' which includes a couple of improvements compared to RC1 (cf. release notes below). A note to our MacOS users: in order to start GraXpert from your applications you should download the .dmg file, copy the contained application to your applications and start it from there. For the first application startup you probably have to allow the execution after your first double click: Go to System Preferences -> Security and Privacy and you should see an entry where you can whitelist GraXpert. The first startup might take a while. Subsequent startups should be considerably faster.

Thanks a lot for your comprehensive and positive feedback so far!

You can download the release here: https://github.com/S...ases/tag/v0.1.3
(we will update the download links at https://www.graxpert.com/ shortly)

Supported Systems

We have built and tested GraXpert on:

• Ubuntu 20.04. (i.e., Debian 11 should work as well)
• Windows 10
• MacOS 10.15 (i.e., any MacOS release afterward should work)

Improvements since RC1

• The grid selection now chooses the darkest quadrant around a grid point
• Added options to adjust the size and color of sample points
• Available key bindings are now documented as part of the help panel
• Reduced perceived noise in stretched previews
• Added a panel 'advanced' to adjust RBF and spline settings, the display language, and other settings
• Improved language detection for German language variants like 'de_CH'
• .fts files are now recognized as fits
• We know read and interpret the 'ROWORDER' field of .fits files to avoid their incorrect, flipped display
• Improved format detection for capitalized file endings (e.g., .TIFF)
• Scrollbars are now scrollable via the mouse wheel
• Incorporated a first, self-compiled PyInstaller bootloader. This should at least reduce the false-positive flagging of GraXpert by some malware detection engines on Windows
• GraXpert is now available as .zip bundles for both Windows and Linux which should improve startup time compared to the self-extracting binary distributions
• Improved the startup time on MacOS

Fixed Bugs

• Fixed a bug where GraXpert refused to start in case it encountered an invalid preferences file
• Fixed a bug where GraXpert refused to start during iteration of available monitors and their dpi

Known Issues

• (bug) GraXpert won't start on some Linux systems with multiple monitors

Full Changelog: https://github.com/S...v0.1.2...v0.1.3


In addition to spline-based interpolation that Siril includes today we provide two additional interpolation methods (RBF and Kriging) that should provide better results in case of complex gradients.
In the meantime we are in contact with the Siril developers and together we are working on an integration of our algorithms into Siril - stay tuned

CS
David

[Steffenhir]

Hi everyone,

 

we have now reached version v0.1.6a (Tethys RC-5a) of GraXpert with several improvements, new features and bug fixes. Among other things we introduced:

• a slider for color saturation (for example to make color gradients and Ha-regions more visible)
• support for Pixinsight's xisf format
• a crop tool (for example to cut off stacking artifacts)
• several bug fixes for Windows and MacOS which delayed or prevented the program from starting

 

You can always find the newest version at https://github.com/S...aXpert/releases

 

CS Steffen

[blakesphere]

I just wanted to say THANK YOU very much.

 

I played with GraXpert on the weekend. Watched Frank's video today and skimmed the user manual.

 

When I apply the gradient removal to the original FITS files, I am getting great results.

[martinclayden]

Just found out about this and thought I'd give it a try. Preliminary indications are that, in terms of removing gradients, it works well - after all that's the prime objective

 

But I do notice that this result seems to be at the expense of additional image noise

 

Have others noticed this ?

 

TIA

 

Martin

[jdmatthew]

I am now using BlurXterminator, NoiseXterminator, StarXterminator and Graxpert.  Can anyone give a reasonable answer as to what order these should be used to work most effectively?  I've been going BlurXTerminator, then Graxpert, then NoiseXterminator and it seems to be working well and cutting out a lot of the steps I used to take to process an image.  However, I am curious as to whether there is an advantage to changing the order in which I'm using these.  I've done a little trial and error and haven't noticed a lot of difference.

 

BTW, it was nice to get one of these for free.  I've paid more for the various ___Xterminator softwares than the original cost of Pixinsight.  Not that they aren't worth it, but it's nice to get a free piece of quality software now and then.

[arbit]

I am now using BlurXterminator, NoiseXterminator, StarXterminator and Graxpert. Can anyone give a reasonable answer as to what order these should be used to work most effectively? I've been going BlurXTerminator, then Graxpert, then NoiseXterminator and it seems to be working well and cutting out a lot of the steps I used to take to process an image. However, I am curious as to whether there is an advantage to changing the order in which I'm using these. I've done a little trial and error and haven't noticed a lot of difference.

BTW, it was nice to get one of these for free. I've paid more for the various ___Xterminator softwares than the original cost of Pixinsight. Not that they aren't worth it, but it's nice to get a free piece of quality software now and then.

General recommendation is : Crop to remove stacking edges > Gradient removal > Color Calibration > BxT (or decon) > other stuff.

You can do a mild noise reduction before stretching. Then star removal, processing and a NR again at the end.

Sent from my SM-S908E using Tapatalk

[Lead_Weight]

Very nice, I just tried it out on my Mac, and it did a great job. I like how when you add points to the grid it moves them to make sure you don't go over a star. Clever.

[view_into_space]

I just released a review and test of this software: https://youtu.be/_K87XIeLf1E