PixInsight New Multiscale Gradient Removal Tool in the Works!!! They need our data contribution to make it happen. Read for details. - Forums

Topic starter

IOTD Staff

Nov 12, 2023 · Edited: Nov 13, 2023

Hello Everyone,

I wanted to bring to your attention an exciting update for those using PixInsight, as highlighted in the email from the PI team this morning.

They are in the process of developing what seems to be a revolutionary tool for gradient removal.For more information about this developing tool, you can read this article: Multiscale Gradient Correction Tutorial. The success of this tool hinges on the completion of "The Multiscale All-Sky Reference Survey (MARS)." More details on this can be found here: MARS Project Documentation.

Here's what the PI team needs from us, the PixInsight user community:

Image Contributions:

They are seeking wide-field images with a field of view between 3 and 50 degrees, particularly from the southern hemisphere.

Specifications for the Images:

Camera Types: Images can be captured using either mono or color cameras, covering RGB bands. For mono cameras, they also request narrowband H-alpha, [O-III], and [S-II] images.
Broadband Images: If you're submitting broadband images, they should be taken under dark skies (Bortle scale 1 to 4).
Preprocessing: Images must be fully preprocessed master lights without any further processing. The team needs the original masters in XISF format, unaltered by any gradient correction. (FITS are also Welcomed)
Drizzle Requirement: All master lights should be processed using the drizzle technique (even at drizzle x1). This is particularly important for color cameras to maintain accurate stellar photometry.

To contribute your data or for further instructions, please visit: MARS Project Contribution Thread. I'm excited to see how our contributions will aid in this groundbreaking development!

Helpful Engaging

Steeve Body

Topic starter

IOTD Staff

Nov 12, 2023

I have been using GraXpert instead of DBE for my past few images, especially their AI interpolation methods for removing gradient which has been working well for me… but this will be for sure the way to do it!

Brian Boyle

Nov 12, 2023

I presume the Astrobin Community Survey data might be useful:

https://www.astrobin.com/forum/c/astrophotography/other/the-astrobin-community-abc-all-sky-survey-announcement/

Although we didn't get a huge take up from the Astrobin community, a few of have been plugging away over the past few months. Progress is recorded here and we are still looking for voiunteers in Bortle 1-4.

https://docs.google.com/spreadsheets/d/1eyHT4dZhSlMvfJHPtBrt93e5wDW6eJbs23bMJ7gjz1o/edit#gid=1598116401

40% of the Southern Sky already

@James Tickner working on survey stitching at the moment so may also have something else to contribute.

Brian

Steeve Body

Topic starter

IOTD Staff

Nov 12, 2023

Brian Boyle:
I presume the Astrobin Community Survey data might be useful:

https://www.astrobin.com/forum/c/astrophotography/other/the-astrobin-community-abc-all-sky-survey-announcement/

Although we didn't get a huge take up from the Astrobin community, a few of have been plugging away over the past few months. Progress is recorded here and we are still looking for voiunteers in Bortle 1-4.

https://docs.google.com/spreadsheets/d/1eyHT4dZhSlMvfJHPtBrt93e5wDW6eJbs23bMJ7gjz1o/edit#gid=1598116401

40% of the Southern Sky already

@James Tickner working on survey stitching at the moment so may also have something else to contribute.

Brian

Most certainly will.. this is pretty much exactly what they are looking for from what I can gather!

Mohammed Khalifa

Nov 13, 2023

Steeve Body:
I have been using GraXpert instead of DBE for my past few images, especially their AI interpolation methods for removing gradient which has been working well for me... but this will be for sure the way to do it!

I have been using GraXpert when they introduced the AI interpolation methods. I tried it on extreme gradients but it wasn’t correcting it enough, there was still some artifacts left in the corners. But on “not extreme” gradients it was working fine and produced completely flat images.

I did read the article of MARS, simply I think it will be better. The method used seems to be more accurate and will produce more natural images. Can’t wait for it !

Alberto Ibañez

Nov 13, 2023 · Edited: Nov 13, 2023

Hi, long time without participating, but I wanted to point out how powerful this old, but generally unknown, tool is, for those that doesn't tried it yet.

GraXpert does a great job with most of the gradients we usually have to deal with, but as for DBE or other tools (as far as I know) they are always neutralizing the background based on where we put our samples (we, or the software), so in the end, there's some interpretation in all of this.

On the other hand, Mulstiscale Gradient Removal takes a "true" reference for calculating the difference between our images and this reference, making a true removal of the unwanted gradient. It doesn't neutralizes our images, it matches them to a "true" reference and this is far better than neutralizing.

Edit: Important to say also that this method has no limitations. Doesn't matter how bad is your gradient. It will remove it completely in one pass.

Helpful

Steeve Body

Topic starter

IOTD Staff

Nov 13, 2023

Alberto Ibañez:
Hi, long time without participating, but I wanted to point out how powerful this old, but generally unknown, tool is, for those that doesn't tried it yet.

GraXpert does a great job with most of the gradients we usually have to deal with, but as for DBE or other tools (as far as I know) they are always neutralizing the background based on where we put our samples (we, or the software), so in the end, there's some interpretation in all of this.

On the other hand, Mulstiscale Gradient Removal takes a "true" reference for calculating the difference between our images and this reference, making a true removal of the unwanted gradient. It doesn't neutralizes our images, it matches them to a "true" reference and this is far better than neutralizing.

Edit: Important to say also that this method has no limitations. Doesn't matter how bad is your gradient. It will remove it completely in one pass.

Indeed! I believe this tools will be a game changer for gradient removal. Living in a bortle 6 zone this will make a great difference for me. I can only imagine how much better mosaics are going to be with this tool as well!

Zak Jones

Nov 13, 2023

I am very interested to see how this new tool works when it is released. It sounds might be the go-to solution for removing gradients, which will definitely help removing extremely complex gradients. I am also eager to see if it will remove gradients from images that haven't been calibrated with flat frames.

Like a few other users have mentioned in this post, I also use GraXpert AI with great results. It just shows how far technology has advanced with the introduction of AI, but this new method sounds like it most likely will make AI gradient removals obsolete.

Props to the PixInsight development team for producing new innovate tools for us astrophotographers to use (and Russ Croman for his innovative tools too)!

Zak

Respectful Supportive

Mohammed Khalifa

Nov 13, 2023

Zak Jones:
I am very interested to see how this new tool works when it is released. It sounds might be the go-to solution for removing gradients, which will definitely help removing extremely complex gradients. I am also eager to see if it will remove gradients from images that haven't been calibrated with flat frames.

Like a few other users have mentioned in this post, I also use GraXpert AI with great results. It just shows how far technology has advanced with the introduction of AI, but this new method sounds like it most likely will make AI gradient removals obsolete.

Props to the PixInsight development team for producing new innovate tools for us astrophotographers to use (and Russ Croman for his innovative tools too)!

Zak

Good point mentioning the possibility to remove gradients with images that are not calibrated with flats, but does it remove dust spots?

Engaging Supportive

Zak Jones

Nov 13, 2023

Zak Jones:
I am very interested to see how this new tool works when it is released. It sounds might be the go-to solution for removing gradients, which will definitely help removing extremely complex gradients. I am also eager to see if it will remove gradients from images that haven't been calibrated with flat frames.

Like a few other users have mentioned in this post, I also use GraXpert AI with great results. It just shows how far technology has advanced with the introduction of AI, but this new method sounds like it most likely will make AI gradient removals obsolete.

Props to the PixInsight development team for producing new innovate tools for us astrophotographers to use (and Russ Croman for his innovative tools too)!

Zak

Good point mentioning the possibility to remove gradients with images that are not calibrated with flats, but does it remove dust spots?

That's why I mentioned it for, as I have used GraXpert AI to remove gradients from my early images where I didn't take flats.

To answer your question, with GraXpert AI, it does not remove any dust spots from my testing. I have to use the Remove Tool later on in Photoshop to remove them.

Zak

Concise

Alberto Ibañez

Nov 14, 2023

@Mohammed Hussain , it will correct large scale features, so in case of a dust spot it probably will left the small scale features of the spot (the edges I guess) so I wouldn't expect too much in this case. This is because in the workflow you use MMT to "blur" both reference image and a copy of the target image, up to 256 scale before subtracting one from the other, so scales below this threshold wouldn't be corrected. This is desirable, to keep your image details. In fact, I usually fine tune the resulting correction model by blurring it in some degree, to improve the "matching".

Helpful