Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Please advise on how to manage a MOSAIC?

  • Please log in to reply
25 replies to this topic

#1 Oscar Szentirmai

Oscar Szentirmai

    Viking 1

  • -----
  • topic starter
  • Posts: 727
  • Joined: 28 Dec 2006
  • Loc: London - UK

Posted 25 January 2023 - 04:01 PM

I'm fairly experienced in imaging, and have 3 years experience, but have never tried a mosaic before.

 

I am currently imaging the Spaghetti Nebula, and have setup NINA to take a 4 panel mosaic with 20% overlap.

 

I usually pre-process my images by adding all my OSC lights, Flat master and dark master in Pixinsight WBPP, and then post process the MASTER LIGHT.

 

How do I go about pre-processing this 4 panel mosaic?

 

I am aiming at 100 300s subs for EACH panel.

 

Master flat is the same for the 4 sets of subs, as the imaging train is untouched between the imaging sessions.

 

How do I go about doing this?

 

Do I process each set separately, and then combine the FOUR MASTER LIGHTS somehow?

 

Or do I calibrate, then register them all together? and if so, how?


Edited by Oscar Szentirmai, 25 January 2023 - 04:37 PM.


#2 italic

italic

    Mariner 2

  • *****
  • Posts: 225
  • Joined: 05 Apr 2022
  • Loc: San Francisco, CA

Posted 25 January 2023 - 04:56 PM

The basic workflow would be to preprocess and stack your panels like normal, then stitch them together with one of the tools in PI. I've tried using Mosaic by Coordinates and Photometric Mosaic scripts, but not Gradient Merge Mosaic. I much prefer Photometric Mosaic because you don't have to punch out stars on the seam that result in pinched stars.

 

I'll outline my general workflow here. I shoot mono, so if you're only OSC, you can ignore the notes about separate channels.

 

My best results came from this general flow:

  • Calibrate raw subs with flats and darks
  • Remove linear gradients from each calibrated sub (can be batched with an image container and ABE process)
  • Register and stack gradient-corrected subs for each panel (not as a whole, only the panel area)
  • Crop out stacking artifacts
  • Final gradient removal on each stack
  • Generate a star field that covers your mosaic's total field of view (with extra) at your imaging scale (I used Catalog Star Generator script)
  • Register each panel to the generated star field
  • Stitch panels into rows, then stitch rows into full image (per channel if mono)
  • Post process like normal

 

This was a lot of work to figure out initially, but now that I have a pretty solid workflow, it actually doesn't take all that long to complete. After stacking, I can get the mosaic put together in an evening (for four channels in LRGB of a six-panel mosaic). OSC would be faster to put together, but computations will take longer for each operation (three channels at a time instead of one). I spend the extra time to tune ABE for each channel of each panel because I wanted the best result before stacking, but I'm sure you could do it roughly and still get a good result.

 

All said, it's a lot of work, but I don't think anybody's making mosaics to save the work. Hell, I don't think any of us are here because it's easy.


  • Oscar Szentirmai likes this

#3 smiller

smiller

    Apollo

  • *****
  • Posts: 1,282
  • Joined: 27 Oct 2018
  • Loc: Vancouver Washington (not BC!)

Posted 25 January 2023 - 07:27 PM

Wow,

 

This looks a lot harder than in APP.   I've been an APP user and just started using PixInsight for a few operations and for blurXterminator..... Mosaics were easy in APP, I've been only doing deep sky for a few months and I've done a half dozen mosaics, up to 6 panels, but I'll be frank...  This sort of scares me from switching to PI for mosaics...

 

I guess I was hoping that if you had good overlap between the panels so they could be referenced to each other, all you had to do is load them all up and say "Stack away you big fancy expensive program!".   ;-)

 

Maybe it's just my old brain wanting to minimize the new things I have to learn when I already have something that works well...

 

Cheers,

 

Steven


  • Oscar Szentirmai likes this

#4 ntph

ntph

    Viking 1

  • -----
  • Posts: 898
  • Joined: 20 Oct 2012
  • Loc: Sudbury, ON

Posted 25 January 2023 - 07:29 PM

Since you are using NINA, and you set up the mosaic in NINA, you should have 4 sets of subs:  Panel 1, 2,3 and 4. Just open WBPP and use the Directory button on the bottom to load your files all at once by picking whatever folder holds the Panels subs. NINA will sort them out for you! Add the appropriate calibration frames, make the settings what you normally would in the Lights tab and the Calibration tab and enter the word Panel in the Grouping Keywords space and check both Pre and Post. Use auto by Panel for registration. You'll get the four panels' master lights in the destination folder. From there,  definitely use the PhotometricMosaic suite in the Scripts menu --it works a charm and has fantastic documentation. John Murphy suggests that if your gradients aren't horribly different, you can wait until the mosaic is assembled before doing background extraction.( Of course he suggests using NormalizeScaleGradient if you go that route!)  He also says it's OK to do BE on each panel, so take your pick depending on how you think your data looks. Certainly simpler and usually effective to do it on the assembled mosaic image in my experience. Use the linear masters for the whole PMM process. It is probably better for you to read the Impatient directions and the full documentation rather than me repeat them and miss something! The end result is one giant masterlight (which it is!)  My experience with using the PMM scripts is that it is actually quite simple, straightforward and actually pretty quick (when you follow the instructions!) and the default settings are appropriate for most projects. When they aren't, the documentation will rescue you. Still having problems? Post it in the PixInsight Forum for PMM and John usually answers pretty quickly. Of course you can try here too. .Jeffery's process above I am sure works, but I have had very good results using the steps above with mosaics up to 4x4. YMMV, but have fun with it. 


Edited by ntph, 25 January 2023 - 07:39 PM.

  • Oscar Szentirmai likes this

#5 Oscar Szentirmai

Oscar Szentirmai

    Viking 1

  • -----
  • topic starter
  • Posts: 727
  • Joined: 28 Dec 2006
  • Loc: London - UK

Posted 25 January 2023 - 07:57 PM

The basic workflow would be to preprocess and stack your panels like normal, then stitch them together with one of the tools in PI. I've tried using Mosaic by Coordinates and Photometric Mosaic scripts, but not Gradient Merge Mosaic. I much prefer Photometric Mosaic because you don't have to punch out stars on the seam that result in pinched stars.

 

I'll outline my general workflow here. I shoot mono, so if you're only OSC, you can ignore the notes about separate channels.

 

My best results came from this general flow:

  • Calibrate raw subs with flats and darks
  • Remove linear gradients from each calibrated sub (can be batched with an image container and ABE process)
  • Register and stack gradient-corrected subs for each panel (not as a whole, only the panel area)
  • Crop out stacking artifacts
  • Final gradient removal on each stack
  • Generate a star field that covers your mosaic's total field of view (with extra) at your imaging scale (I used Catalog Star Generator script)
  • Register each panel to the generated star field
  • Stitch panels into rows, then stitch rows into full image (per channel if mono)
  • Post process like normal

 

This was a lot of work to figure out initially, but now that I have a pretty solid workflow, it actually doesn't take all that long to complete. After stacking, I can get the mosaic put together in an evening (for four channels in LRGB of a six-panel mosaic). OSC would be faster to put together, but computations will take longer for each operation (three channels at a time instead of one). I spend the extra time to tune ABE for each channel of each panel because I wanted the best result before stacking, but I'm sure you could do it roughly and still get a good result.

 

All said, it's a lot of work, but I don't think anybody's making mosaics to save the work. Hell, I don't think any of us are here because it's easy.

Many thanks for your reply...

 

Ok, that part where I said "I'm fairly experienced in imaging".....

 

:)

 

I take that back....

 

Do you mind explaining that a little bit simpler?

 

I'm sorry, but I do not follow.

 

:)



#6 Oscar Szentirmai

Oscar Szentirmai

    Viking 1

  • -----
  • topic starter
  • Posts: 727
  • Joined: 28 Dec 2006
  • Loc: London - UK

Posted 25 January 2023 - 08:04 PM

Since you are using NINA, and you set up the mosaic in NINA, you should have 4 sets of subs:  Panel 1, 2,3 and 4. Just open WBPP and use the Directory button on the bottom to load your files all at once by picking whatever folder holds the Panels subs. NINA will sort them out for you! Add the appropriate calibration frames, make the settings what you normally would in the Lights tab and the Calibration tab and enter the word Panel in the Grouping Keywords space and check both Pre and Post. Use auto by Panel for registration. You'll get the four panels' master lights in the destination folder. From there,  definitely use the PhotometricMosaic suite in the Scripts menu --it works a charm and has fantastic documentation. John Murphy suggests that if your gradients aren't horribly different, you can wait until the mosaic is assembled before doing background extraction.( Of course he suggests using NormalizeScaleGradient if you go that route!)  He also says it's OK to do BE on each panel, so take your pick depending on how you think your data looks. Certainly simpler and usually effective to do it on the assembled mosaic image in my experience. Use the linear masters for the whole PMM process. It is probably better for you to read the Impatient directions and the full documentation rather than me repeat them and miss something! The end result is one giant masterlight (which it is!)  My experience with using the PMM scripts is that it is actually quite simple, straightforward and actually pretty quick (when you follow the instructions!) and the default settings are appropriate for most projects. When they aren't, the documentation will rescue you. Still having problems? Post it in the PixInsight Forum for PMM and John usually answers pretty quickly. Of course you can try here too. .Jeffery's process above I am sure works, but I have had very good results using the steps above with mosaics up to 4x4. YMMV, but have fun with it. 

Many thanks

 

That seems simpler!

 

Let me get this right..

 

You mentioned: NINA will sort them out for you!

 

Didn't you mean PixInsight will sort them for you?

 

So, I open WBPP as I always do.

 

In the LIGHTS tab, I will add ALL the subs from the 4 panels, then add Master Flat, then Master Dark.

 

But how will WBPP know which is which?

 

I suppose that is where the KEYWORD you mentioned comes in?

 

Never used the KEYWORD box before.

 

I will look into the documentation you mentioned, but can you please elaborate on the KEYWORD bit, and how it grouped the LIGHTS into different groups?



#7 italic

italic

    Mariner 2

  • *****
  • Posts: 225
  • Joined: 05 Apr 2022
  • Loc: San Francisco, CA

Posted 25 January 2023 - 08:44 PM

Many thanks for your reply...

 

Ok, that part where I said "I'm fairly experienced in imaging".....

 

smile.gif

 

I take that back....

 

Do you mind explaining that a little bit simpler?

 

I'm sorry, but I do not follow.

 

smile.gif

Very simply, stack each panel like you normally would, then stitch them into the mosaic. I came to the workflow I did because I redid the project half a dozen times and worked out all the problems I had in the captured data. Probably the biggest help in getting everything stitched together was registering against a generated star field. All the panels will be aligned and undistorted, and they should all fit together nicely.


  • Oscar Szentirmai likes this

#8 CorralesRay

CorralesRay

    Mariner 2

  • *****
  • Posts: 233
  • Joined: 13 Jun 2018

Posted 25 January 2023 - 09:22 PM

Watch "The Astro Imaging Channel's" video on YouTube: https://youtu.be/urcTP3IMg1I 

 

this provided me with a very useful process and my Mosaic turned out well.

 

Best of luck,

 

Ray


  • Oscar Szentirmai likes this

#9 TelescopeGreg

TelescopeGreg

    Aurora

  • -----
  • Posts: 4,801
  • Joined: 16 Jul 2018
  • Loc: Auburn, California, USA

Posted 25 January 2023 - 10:05 PM

I've only done one mosaic (so I'm an expert, right?), but it turned out to be really easy.  As italic said, stack each of the panels (this was a 2x2 mosaic of the Heart Nebula) individually, then stitch them together.  ASTAP did both operations, normal stacking for the panels, then using "Image stitching mode" and "Astrometric alignment" to do the stitching.  Remember to only use the calibration frames (Darks, Flats) on the individual frames.  Process the resulting combined fits file (warning, it will be large!) as normal. 


  • Oscar Szentirmai likes this

#10 ntph

ntph

    Viking 1

  • -----
  • Posts: 898
  • Joined: 20 Oct 2012
  • Loc: Sudbury, ON

Posted 25 January 2023 - 10:31 PM

Many thanks

 

That seems simpler!

 

Let me get this right..

 

You mentioned: NINA will sort them out for you!

 

Didn't you mean PixInsight will sort them for you?

 

So, I open WBPP as I always do.

 

In the LIGHTS tab, I will add ALL the subs from the 4 panels, then add Master Flat, then Master Dark.

 

But how will WBPP know which is which?

 

I suppose that is where the KEYWORD you mentioned comes in?

 

Never used the KEYWORD box before.

 

I will look into the documentation you mentioned, but can you please elaborate on the KEYWORD bit, and how it grouped the LIGHTS into different groups?

My mistake--you're right--PI WBPP sorts it out. NINA sets the stage though, if you used it to plan your mosaic panels. In the folder where NINA puts your subs, you will see the file names will include "Panel # whatever). The keyword that you want  PI to use will be Panel. So in WBPP, on the right, you see Grouping Keywords. Click in the top box and type Panel then click on the big green plus beside it. That will put PANEL in the box beneath and a check under Pre. Now all your files will be sorted and if you just go from there, with all the normal weighting, registration and integration, you will have... a mess. You need to add a Keyword for Post as well. Simple--click on the line that says Panel (under Keyword-Pre-Post) and the line turns orange. Now click on the green and black gear symbol above the trashcan and toggle the checkmarks. You need a check under both Pre and Post. Now if you look under the Post-Calibration tab across the top, you will see all the panels, the number of files and total integration times for each panel. Magic!   Wasn't that simple and totally intuitive??!! I guess the PI team thinks so. lol.gif   The last thing is to select "auto by PANEL " under Registration Reference Image, bottom right of WBPP.

 

The big thing is your file names have to have Panel in them, which they will if you use NINA for the whole thing. The calibration frames don't need anything other than your usual naming and loading. Hope that helps. Here are a few screen shots showing what things should look like. 

 

Hope this helps and that I have explained things correctly. 

 

I really really suggest you read the documentation for the PMM suite. It is clear, concise and will get you through with a minimum of fuss and bother. OK, there are a few steps to go through, but do it once and you'll say, "Well that wasn't so bad and look at the great result." Do it again and you say, "Hey I think I am getting to like this" and from then on, it's " well, this is going to be easy peasy".   I know there are lots of other mosaic-skinning programs out there, some that may be easier or faster, but I will skin my mosaic cats with PMM.  Tried GMM on my first one, got pinched stars, wonky seams. Same data first crack with PMM was perfect. What more can I say--I am a fan. 

med_gallery_213329_17534_29693.png

med_gallery_213329_17534_74009.png


Edited by ntph, 25 January 2023 - 10:41 PM.

  • Oscar Szentirmai likes this

#11 Yerman

Yerman

    Mariner 2

  • -----
  • Posts: 249
  • Joined: 20 Apr 2022
  • Loc: Ontario, Canada

Posted 25 January 2023 - 10:41 PM

It’s actually much simpler that any of the methods mentioned here. There is a Microsoft program I think it’s called ICE sorry I’m not at my computer to verify the name. You just down load it and it’s basically four steps and your done. It’s fully automatic you just have to tell it the order that the images were taken. In a four panel mosaic you put in the four already stacked images. You don’t even need to calibrate them in fact I think it prefers if they are not calibrated. It will colour balance the mosaic and even out the gradients. If need be it will stretch or shrink the edges to ensure that the panels line up properly. After it’s finished you can process the image as you see fit. I was amazed at how it balanced out the four gradients and differing intensities especially for things like if the moon comes up during a long shoot or some of the images are getting closer to the horizon where it’s brighter. 


  • Oscar Szentirmai likes this

#12 Oscar Szentirmai

Oscar Szentirmai

    Viking 1

  • -----
  • topic starter
  • Posts: 727
  • Joined: 28 Dec 2006
  • Loc: London - UK

Posted 26 January 2023 - 06:05 AM

My mistake--you're right--PI WBPP sorts it out. NINA sets the stage though, if you used it to plan your mosaic panels. In the folder where NINA puts your subs, you will see the file names will include "Panel # whatever). The keyword that you want  PI to use will be Panel. So in WBPP, on the right, you see Grouping Keywords. Click in the top box and type Panel then click on the big green plus beside it. That will put PANEL in the box beneath and a check under Pre. Now all your files will be sorted and if you just go from there, with all the normal weighting, registration and integration, you will have... a mess. You need to add a Keyword for Post as well. Simple--click on the line that says Panel (under Keyword-Pre-Post) and the line turns orange. Now click on the green and black gear symbol above the trashcan and toggle the checkmarks. You need a check under both Pre and Post. Now if you look under the Post-Calibration tab across the top, you will see all the panels, the number of files and total integration times for each panel. Magic!   Wasn't that simple and totally intuitive??!! I guess the PI team thinks so. lol.gif   The last thing is to select "auto by PANEL " under Registration Reference Image, bottom right of WBPP.

 

The big thing is your file names have to have Panel in them, which they will if you use NINA for the whole thing. The calibration frames don't need anything other than your usual naming and loading. Hope that helps. Here are a few screen shots showing what things should look like. 

 

Hope this helps and that I have explained things correctly. 

 

I really really suggest you read the documentation for the PMM suite. It is clear, concise and will get you through with a minimum of fuss and bother. OK, there are a few steps to go through, but do it once and you'll say, "Well that wasn't so bad and look at the great result." Do it again and you say, "Hey I think I am getting to like this" and from then on, it's " well, this is going to be easy peasy".   I know there are lots of other mosaic-skinning programs out there, some that may be easier or faster, but I will skin my mosaic cats with PMM.  Tried GMM on my first one, got pinched stars, wonky seams. Same data first crack with PMM was perfect. What more can I say--I am a fan. 

med_gallery_213329_17534_29693.png

med_gallery_213329_17534_74009.png

Thank you so much for the clarificarion.

 

That is very clear.

 

The only issue though is that for some reason, NINA did not name my subs differently?

 

All my subs are named as any other imaging sub I regularly do, but there is no PANEL in the sub's name?

 

There must be a setting in NINA that I did not set correctly?

 

My subs are in 4 different FOLDERS that were named PANEL1, PANEL2 etc by NINA, but the actual subs inside the folder do not have PANEL in the name?

 

This is what the sub's name looks like:

 

2023-01-25_20-15-16__-4.90_300.00s_0000

 

That was a sub in folder: PANEL3

 

Can I ADD SH2-240 Panel# to each sub, to look like yours, or it should have been set in NINA, and its too late to change them now?



#13 rgenier

rgenier

    Vostok 1

  • -----
  • Posts: 150
  • Joined: 11 Nov 2020
  • Loc: Ontario

Posted 26 January 2023 - 09:37 AM

Thank you so much for the clarificarion.

 

That is very clear.

 

The only issue though is that for some reason, NINA did not name my subs differently?

 

All my subs are named as any other imaging sub I regularly do, but there is no PANEL in the sub's name?

 

There must be a setting in NINA that I did not set correctly?

 

My subs are in 4 different FOLDERS that were named PANEL1, PANEL2 etc by NINA, but the actual subs inside the folder do not have PANEL in the name?

 

This is what the sub's name looks like:

 

2023-01-25_20-15-16__-4.90_300.00s_0000

 

That was a sub in folder: PANEL3

 

Can I ADD SH2-240 Panel# to each sub, to look like yours, or it should have been set in NINA, and its too late to change them now?

I believe NINA will add the Panel label to the FITS header. It shouldn't matter that the PANEL is not listed in your file name. If you set up WBPP with the PANEL grouping keyword, it should sort all of the images.


  • Oscar Szentirmai likes this

#14 ntph

ntph

    Viking 1

  • -----
  • Posts: 898
  • Joined: 20 Oct 2012
  • Loc: Sudbury, ON

Posted 26 January 2023 - 09:40 AM

Oscar, I apologize, I realized when I woke up this morning that I assumed that you put your target name in your subs as I have. . Answering your last questioin first:  I don't know if there is an easy way to change the file names or if you would have to do it one at a time.If there isn't an easy way to batch change  the name, then I think you'd be best to just make your four masters like normal and go from there. But going forward, here is my suggestion:  

Go into the Options-Imaging tab in NINA. Add $$TargetName$$ to your Image File Pattern string and you should be good. 

From then on, you should be good. Not sure but I think you can put TargetName whereever you like in the file name and it will still work. 

 

med_gallery_213329_17534_28500.png


  • Oscar Szentirmai likes this

#15 rgenier

rgenier

    Vostok 1

  • -----
  • Posts: 150
  • Joined: 11 Nov 2020
  • Loc: Ontario

Posted 26 January 2023 - 09:45 AM

My mosaic workflow using an OSC camera:

 

1. Stack each of the panels (using WBPP and the panel keyword).

2. DBE or ABE on each of the stacked images to remove gradients.

3. ImageSolver Script on each Panel (this will add the RA/DEC coordinates of the image to each header).

4. MosaicByCoordinates (this will generate 4x output files - 4 new panels - that will be mostly black with the panel image in a corner. These are the new images that you are going to merge into the mosaic).

5. DNALinearFit Script in order to equalize the brightness between your panels.

6. PhotometricMosaic Script to combine your images into the final mosaic.

7. DBE on the mosaic to remove any remaining gradients.

8. Continue on with your normal processing...



#16 rgenier

rgenier

    Vostok 1

  • -----
  • Posts: 150
  • Joined: 11 Nov 2020
  • Loc: Ontario

Posted 26 January 2023 - 09:54 AM

Confirmed that WBPP will sort by PANEL without it being listed in the file name.

 

 

Attached Thumbnails

  • Screenshot 2023-01-26 095146.jpg
  • Screenshot 2023-01-26 095310.jpg

  • Oscar Szentirmai likes this

#17 Oscar Szentirmai

Oscar Szentirmai

    Viking 1

  • -----
  • topic starter
  • Posts: 727
  • Joined: 28 Dec 2006
  • Loc: London - UK

Posted 26 January 2023 - 10:11 AM

Confirmed that WBPP will sort by PANEL without it being listed in the file name.

I have tried WBPP, and added the LIGHTS from the 4 folders named Panel1, Panel2, Panel3 & Panel4, but WBPP did not segregate them into 4 groups.

They all show as ONE group of lights



#18 rgenier

rgenier

    Vostok 1

  • -----
  • Posts: 150
  • Joined: 11 Nov 2020
  • Loc: Ontario

Posted 26 January 2023 - 12:58 PM

I have tried WBPP, and added the LIGHTS from the 4 folders named Panel1, Panel2, Panel3 & Panel4, but WBPP did not segregate them into 4 groups.

They all show as ONE group of lights

Did you add the grouping keyword in WBPP?


  • Oscar Szentirmai likes this

#19 Oscar Szentirmai

Oscar Szentirmai

    Viking 1

  • -----
  • topic starter
  • Posts: 727
  • Joined: 28 Dec 2006
  • Loc: London - UK

Posted 26 January 2023 - 01:05 PM

Did you add the grouping keyword in WBPP?

Yes

I wrote PANEL, and chose Pre and Post

Yet when I add the LIGHTS from each foler, they are all grouped into one set of lights, with no differentiation


Edited by Oscar Szentirmai, 26 January 2023 - 01:05 PM.


#20 hollo

hollo

    Mariner 2

  • -----
  • Posts: 232
  • Joined: 08 Jan 2022
  • Loc: Hertfordshire, UK

Posted 26 January 2023 - 01:32 PM

I've been playing with this a bit recently, and I think the key messages for WBPP in PixInsight are:

 

1. Pixinsight will look for KEYWORD_VALUE to appear in the path of any image. This can be in a filename, or in a directory. So if you put all of panel 1 into a directory called Panel_p1 it will work just as well as if the subs are called Mybigmosaic_Panel_p1_OIII_60s_009.fits.

2. You add the grouping keywords in the lights tab of WBPP. They add by default as "Pre", but clicking the option button changes the selected one to "Pre and Post", and then "Post".

    - Pre will use different calibration frames. So if you have 2 sessions with different flats then you might have them labelled as night_1 and night_2, and night would be a pre grouping keyword.

    - Post will use the same calibration frames, but be stacked separately. If you took all your subs on the same night with the same flats, then panel_1 and panel_2 will be stacked separately.

    - Both does both (so if you did panel 2 on a second night and it has different calibration frames then you can do this).

    - If you took multiple panels over multiple nights with lots of calibration frames then you will need a night_ and a panel_ grouping, and getting capture software to name appropriately really helps.

3. By default all stacks will use the same reference frame for registration ("auto"). After you've added your Post grouping keyword you almost certainly want to change this to "Auto by PANEL" or whatever grouping keyword you've gone with.

 

After that it does seem to just work. I go with a directory structure of eg:

 

m8182/
├─ lights/
│  ├─ session_20230125/
│  │  ├─ panel_m81/
│  │  ├─ panel_m82/
│  ├─ session_20230126/
│  │  ├─ panel_m81/
│  │  ├─ panel_m82/
├─ flats/
│  ├─ session_20230126/
│  ├─ session_20230125/


  • Oscar Szentirmai likes this

#21 Oscar Szentirmai

Oscar Szentirmai

    Viking 1

  • -----
  • topic starter
  • Posts: 727
  • Joined: 28 Dec 2006
  • Loc: London - UK

Posted 26 January 2023 - 01:40 PM

Did you add the grouping keyword in WBPP?

 

I've been playing with this a bit recently, and I think the key messages for WBPP in PixInsight are:

 

1. Pixinsight will look for KEYWORD_VALUE to appear in the path of any image. This can be in a filename, or in a directory. So if you put all of panel 1 into a directory called Panel_p1 it will work just as well as if the subs are called Mybigmosaic_Panel_p1_OIII_60s_009.fits.

2. You add the grouping keywords in the lights tab of WBPP. They add by default as "Pre", but clicking the option button changes the selected one to "Pre and Post", and then "Post".

    - Pre will use different calibration frames. So if you have 2 sessions with different flats then you might have them labelled as night_1 and night_2, and night would be a pre grouping keyword.

    - Post will use the same calibration frames, but be stacked separately. If you took all your subs on the same night with the same flats, then panel_1 and panel_2 will be stacked separately.

    - Both does both (so if you did panel 2 on a second night and it has different calibration frames then you can do this).

    - If you took multiple panels over multiple nights with lots of calibration frames then you will need a night_ and a panel_ grouping, and getting capture software to name appropriately really helps.

3. By default all stacks will use the same reference frame for registration ("auto"). After you've added your Post grouping keyword you almost certainly want to change this to "Auto by PANEL" or whatever grouping keyword you've gone with.

 

After that it does seem to just work. I go with a directory structure of eg:

 

m8182/
├─ lights/
│  ├─ session_20230125/
│  │  ├─ panel_m81/
│  │  ├─ panel_m82/
│  ├─ session_20230126/
│  │  ├─ panel_m81/
│  │  ├─ panel_m82/
├─ flats/
│  ├─ session_20230126/
│  ├─ session_20230125/

Apologies

 

It seems to have worked.

 

I had named the folders PANEL1 with no space.

 

I changed it to PANEL 1 with a space before the 1, and WBPP detected it!!


  • rgenier likes this

#22 Oscar Szentirmai

Oscar Szentirmai

    Viking 1

  • -----
  • topic starter
  • Posts: 727
  • Joined: 28 Dec 2006
  • Loc: London - UK

Posted 27 January 2023 - 04:19 AM

Oscar, I apologize, I realized when I woke up this morning that I assumed that you put your target name in your subs as I have. . Answering your last questioin first:  I don't know if there is an easy way to change the file names or if you would have to do it one at a time.If there isn't an easy way to batch change  the name, then I think you'd be best to just make your four masters like normal and go from there. But going forward, here is my suggestion:  

Go into the Options-Imaging tab in NINA. Add $$TargetName$$ to your Image File Pattern string and you should be good. 

From then on, you should be good. Not sure but I think you can put TargetName whereever you like in the file name and it will still work. 

 

med_gallery_213329_17534_28500.png

Quick question..

 

I suppose since this is a MOSAIC, I should avoid the AUTOCROP feature in WBPP?

 

In WBPP when I normally process my images, I pick the MASTER_LIGHT_AUTOCROP file, as it clears up the edges for me, but in this case, being a MOSAIC, I suppose I should avoid that?



#23 Oscar Szentirmai

Oscar Szentirmai

    Viking 1

  • -----
  • topic starter
  • Posts: 727
  • Joined: 28 Dec 2006
  • Loc: London - UK

Posted 27 January 2023 - 06:55 AM

In the documentation of PMM, it says:

 

TARGET and REFERENCE images should be registered to each other.

 

Should I assume that since I used WBPP, with the keyword PANEL, and in  REGISTRATION REFERENCE IMAGE box I chose AUTO BY PANEL, that my 4 resultant MASTER LIGHTS are all registrerd to each other?



#24 ntph

ntph

    Viking 1

  • -----
  • Posts: 898
  • Joined: 20 Oct 2012
  • Loc: Sudbury, ON

Posted 27 January 2023 - 10:11 AM

Ah , Oscar, go back and read the documentation for PMM again.lol.gif  Sections 1.1 through 1.3. Nearly everything you need to know is there! You have to  platesolve your panels--WBPP may already have done that for you, but  any cropping will remove the solution--not sure if Autocrop does or not; I don't use it. Add the platesolved panels in MosaicByCoordinates--it generates the mosaic panels with a "registered" tag in the name. This "registers" each panel to its appropriate position in the final mosaic ( top left, top right, bottom left, bottom right) by the plate solution coordinates. Each of these if you apply an STF will look like a normal frame with a large amount of black; each one will be different. Then use TrmMosaicTiles script. Then use PMM:  pick two tiles that share edges, say Panel 1 and Panel 2.( In a 2x2 mosaic you could use the 1-2 and 3-4 pairs or the 1-3 and 2-4; just don't use panels that overlap diagonally)  It doesn't matter which one is target and which is reference. You need to tell it whether to replace the target image or to create a new image, and enter some image scale information--those boxes are right below where you add the reference and target images. For your first try, just leave everything else as is for now. Hit Run. Pmm will generate a new image (or replace the target, depending on what you did above). Look at it with an STF--you now have two panels stitched togetther. Now do the same for the remaining two panels. Now you have two new panels, each with half of the mosaic. Add one of these as target, one as reference and hit Run.  Done. Your first mosaic. 

 

Now go back and look at what some of the buttons do. The ones you will be most likely interested in are in the Mosaic Jon  and Gradient Correction sections. Mouseovers give more information on most things. Hit the Join button to generate the area where the join actually happens. You can adjust the actual line; read the long version dcoumentation. Same thing with the Gradient Correction section. You can see how PMM will deal with the gradient in each of the reference and target images and can make adjustments; again, read the long version of the documentation. 

 

It does take a bit of work, maybe more than some other programs, but it gives you options for control that can make a big difference in how things turn out. Once you've done it once or twice, and you've read the documentation a bit more carefully and understand what it's doing because you've now seen how things work, it becomes pretty straightforward, I thnk.  Good luck and looking forward to seeing what you get! 


Edited by ntph, 27 January 2023 - 10:15 AM.

  • Oscar Szentirmai likes this

#25 Oscar Szentirmai

Oscar Szentirmai

    Viking 1

  • -----
  • topic starter
  • Posts: 727
  • Joined: 28 Dec 2006
  • Loc: London - UK

Posted 27 January 2023 - 01:21 PM

Ah , Oscar, go back and read the documentation for PMM again.lol.gif  Sections 1.1 through 1.3. Nearly everything you need to know is there! You have to  platesolve your panels--WBPP may already have done that for you, but  any cropping will remove the solution--not sure if Autocrop does or not; I don't use it. Add the platesolved panels in MosaicByCoordinates--it generates the mosaic panels with a "registered" tag in the name. This "registers" each panel to its appropriate position in the final mosaic ( top left, top right, bottom left, bottom right) by the plate solution coordinates. Each of these if you apply an STF will look like a normal frame with a large amount of black; each one will be different. Then use TrmMosaicTiles script. Then use PMM:  pick two tiles that share edges, say Panel 1 and Panel 2.( In a 2x2 mosaic you could use the 1-2 and 3-4 pairs or the 1-3 and 2-4; just don't use panels that overlap diagonally)  It doesn't matter which one is target and which is reference. You need to tell it whether to replace the target image or to create a new image, and enter some image scale information--those boxes are right below where you add the reference and target images. For your first try, just leave everything else as is for now. Hit Run. Pmm will generate a new image (or replace the target, depending on what you did above). Look at it with an STF--you now have two panels stitched togetther. Now do the same for the remaining two panels. Now you have two new panels, each with half of the mosaic. Add one of these as target, one as reference and hit Run.  Done. Your first mosaic. 

 

Now go back and look at what some of the buttons do. The ones you will be most likely interested in are in the Mosaic Jon  and Gradient Correction sections. Mouseovers give more information on most things. Hit the Join button to generate the area where the join actually happens. You can adjust the actual line; read the long version dcoumentation. Same thing with the Gradient Correction section. You can see how PMM will deal with the gradient in each of the reference and target images and can make adjustments; again, read the long version of the documentation. 

 

It does take a bit of work, maybe more than some other programs, but it gives you options for control that can make a big difference in how things turn out. Once you've done it once or twice, and you've read the documentation a bit more carefully and understand what it's doing because you've now seen how things work, it becomes pretty straightforward, I thnk.  Good luck and looking forward to seeing what you get! 

Thank you SO much for your detailed explanation. 
 

Earlier today morning, I did actually figure out that I need to do the Mosaic coordinates and trim first. 
 

I printed out the documentation and followed it to the letter. 
 

I think the documentation should make it clearer that BEFORE opening the actual mosaic script, those two things need to be done first. The documentation goes through the process, but did not stress on those two points, or maybe I just did not get it myself. 
 

Anyways, it worked!

 

Its still early days, as I have only 5 hours on panels 1,3 &4 and just 1 hour on panel two. 
 

My skies are not great, and I believe I need much much more integration time for this impossibly dim target. 
 

The result is not great at the joins. I think I need to understand more the options where the graphs show the gradients for RGB. 

 

Inspite of my graphs being almost straight, with no peaks or troughs, I still have significantly visible areas between the 4 panels. 
 

I’ll attach a picture of the preliminary result when I’m back home. 
 

Thanks again for all your help. It was a lot of fun to try this mosaic!


  • ntph likes this


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics