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

Combining OSC Data and Duo-band Data in Pixinsight

  • Please log in to reply
8 replies to this topic

#1 Juno18

Juno18

    Mariner 2

  • *****
  • topic starter
  • Posts: 287
  • Joined: 07 Dec 2018
  • Loc: Long Beach, Mississippi, USA

Posted 12 May 2021 - 08:46 AM

Hi forum,

 

I recently imaged M101 (https://www.cloudyni...9252-m101-help/) and with a little help (thanks!), managed a decent image.

Last week I imaged the same target with an L-eNhance filter to hopefully enhance the galaxy's Ha areas. I had no idea how to use or process this data, but I thought that I would figure it out or find something on Google help me.

 

The imaging went well, but processing didn't. I had a few leads (from the PI forum) and searching on the internet, but I guess that I am just not there (yet). I kept running into steps that didn’t apply or I interpreted them wrong. I only have rudimentary skills in PI.

 

So,I decided to try simple pixelmath to try different combinations of the OSC image and Ha image. Worked easily, but never could get a combination image that looked any better than the original OSC image.

I went out on my own and came up with this idea. Basically, split up each master into individual R, G, and B channels and combine the Ha channel with the OSC green and blue channels.

 

Attempted Workflow: 

 

Open both the OSC and Ha masters
Star align both images (register the Ha image to the OSC image
Run ABE or DBE to remove background and gradients from both images
Split the channels of both the OSC and Ha images
Linear fit both sets of data to their respective green channel
Set the image size (Resample Process) the same for both sets of three channels.
Channel Combination using the Ha_R, OSC_G, and OSC_B

Here is when I got the error.

 

Capture.JPG

 

I really don't understand the error when this is basically the same process that I use with one image and it works fine.

 

I tried the same combination using the process LRGBCombination and got the same error.

 

Of course, this is not a correct approach!

 

Is this something that I should continue to pursue considering my basic skills in PI, or just shuck the data and stick with one filter imaging. I have tried other workflows (like https://www.lightvor...d.html#Section1) and have gotten stuck with not knowing enough to fill in the blanks.

 

Thanks for any guidance that you can give me (including the shucking idea!).  



#2 bulrichl

bulrichl

    Ranger 4

  • *****
  • Posts: 360
  • Joined: 27 May 2018
  • Loc: La Palma (Canary Islands)

Posted 12 May 2021 - 09:41 AM

In order to combine channels, you have to click on the blue sphere icon (Apply global), NOT the blue square icon (Apply).

 

Bernd


  • Juno18 and limeyx like this

#3 jonnybravo0311

jonnybravo0311

    Surveyor 1

  • *****
  • Posts: 1,706
  • Joined: 05 Nov 2020
  • Loc: NJ, US

Posted 12 May 2021 - 10:17 AM

You're on the right track.

 

The first thing to do is to register all of your images to a single reference. This is pretty easy since you've only got 2 images: OSC and OSC+filter. Use the "StarAlign" process. Pick one as the reference and the other as the image to register to that reference. Click the blue circle and you'll get a new image named something like "OSC_registered". This avoids the need to do the image size / resample thing you mentioned and guarantees the images will line up nicely together.

 

Now you'd apply the same dynamic crop to both to get the same framing.

 

Here's where you'd split things. Quite honestly, you don't really need to split the OSC image into its components. You can keep it as the RGB image. The image you _do_ want to split is the one you got from using the dual band filter. The R channel becomes Ha. If you want, you can use pixel math to produce an image that is the combination of the G and B channels and call it O3. (Try something like max(G,B) as your math).

 

There's this neat script called NBRGBCombination. You can play around with all different kinds of combinations with it. You use your RGB image, then add the R channel from the dual band filter image as Ha.


  • Juno18 and limeyx like this

#4 Juno18

Juno18

    Mariner 2

  • *****
  • topic starter
  • Posts: 287
  • Joined: 07 Dec 2018
  • Loc: Long Beach, Mississippi, USA

Posted 12 May 2021 - 10:50 AM

Thanks Bernd and Johnny!

 

Johnny, I am at the NBRGBCombination Process. Do I select the RGB image as the L channel and the Ha_R as the R channel?  For some reason, the RGB image is not listed as a source image. I will have to start over again. I must have got PI (and myself) confused.

 

I will try to start fresh this afternoon. Johnny, if you can point me to what files to use for the LRGBComb Process just for a basic OSC+Ha_R combination. I will try again later using just the Ha_l and the OSC image. If I have success, I will try the Pixelmath G and B combination. 

 

Thanks you both for your help. I realize that I am challenged by this! Good learnings though!

 

NOTE: the two masterlights are inverted. Should I change the Geometry in one to make them match?


Edited by Juno18, 12 May 2021 - 10:53 AM.


#5 Peregrinatum

Peregrinatum

    Surveyor 1

  • *****
  • Posts: 1,733
  • Joined: 27 Dec 2018
  • Loc: South Central Valley, Ca

Posted 12 May 2021 - 11:18 AM

This is the way I like to do this, it is an iterative process to taste:

Split all images into R, G, and B, making sure they are all registered to one another.

In PixelMath place in the R, G, B images from the OSC.

Starting with the NB R image combine it with the OSC R image in two different ways, [1] weighted average, [2] max pixel value, here is an example of each:

[1] R*0.6 + Ha*0.4

[2] max(R, Ha), you can also weight within the max function: max(R, 1.2*Ha)

Compare the new image to the the base line OSC image and look for differences that you like, you can do this by toggling one image over the other,

To toggle images place one image on top of the other, then hit <Ctrl><PgDn> and it will blink the images to see the effects.

Once you find a new image that you like, this becomes the new base line image.

Now rinse and repeat for the G channel using whatever NB images you choose, then for the B channel.

 

Takes time, but if you are patient it will yield very good results.


Edited by Peregrinatum, 12 May 2021 - 11:19 AM.

  • Juno18 likes this

#6 jonnybravo0311

jonnybravo0311

    Surveyor 1

  • *****
  • Posts: 1,706
  • Joined: 05 Nov 2020
  • Loc: NJ, US

Posted 12 May 2021 - 12:31 PM

Thanks Bernd and Johnny!

 

Johnny, I am at the NBRGBCombination Process. Do I select the RGB image as the L channel and the Ha_R as the R channel?  For some reason, the RGB image is not listed as a source image. I will have to start over again. I must have got PI (and myself) confused.

 

I will try to start fresh this afternoon. Johnny, if you can point me to what files to use for the LRGBComb Process just for a basic OSC+Ha_R combination. I will try again later using just the Ha_l and the OSC image. If I have success, I will try the Pixelmath G and B combination. 

 

Thanks you both for your help. I realize that I am challenged by this! Good learnings though!

 

NOTE: the two masterlights are inverted. Should I change the Geometry in one to make them match?

OK... you should have 2 images loaded into PI. Image 1 = the OSC image. Image 2 = the dual-band filter image. Both of these are color images (i.e. shot on the same camera, just that in one case you used some dual band filter like the Optolong L-eNhance), correct?

 

Step 1: register image 2 to image 1. This will take care of your inversion as well. Creates image2_registered.

Step 2: the same dynamic crop on both images. Draw the box on one image. Drag triangle to the other image, then click green checkbox.

Step 3: Split image2_registered into components (R, G, B)

Step 4: Open NBRGBCombination script. Under "RGB source Image" choose image 1. Under "Narrowband for R" choose image2_registered_R

Step 5: Play with different values for bandwidth and scale and click NBRGB button

Step 6: When you find one that makes you happy, click magic OK button

Step 7: You now have a new image called NBRGBCombination


  • Juno18 and limeyx like this

#7 Juno18

Juno18

    Mariner 2

  • *****
  • topic starter
  • Posts: 287
  • Joined: 07 Dec 2018
  • Loc: Long Beach, Mississippi, USA

Posted 12 May 2021 - 01:36 PM

Thanks Peregrinatum and Johnny for information!

 

I can't wait to get into PI and try out. I have a few things going on this afternoon and hopefully can give these a try this evening. 

 

Thanks for the quick help! I will post back soon.



#8 Juno18

Juno18

    Mariner 2

  • *****
  • topic starter
  • Posts: 287
  • Joined: 07 Dec 2018
  • Loc: Long Beach, Mississippi, USA

Posted 12 May 2021 - 03:38 PM

OK... you should have 2 images loaded into PI. Image 1 = the OSC image. Image 2 = the dual-band filter image. Both of these are color images (i.e. shot on the same camera, just that in one case you used some dual band filter like the Optolong L-eNhance), correct?

 

Step 1: register image 2 to image 1. This will take care of your inversion as well. Creates image2_registered.

Step 2: the same dynamic crop on both images. Draw the box on one image. Drag triangle to the other image, then click green checkbox.

Step 3: Split image2_registered into components (R, G, B)

Step 4: Open NBRGBCombination script. Under "RGB source Image" choose image 1. Under "Narrowband for R" choose image2_registered_R

Step 5: Play with different values for bandwidth and scale and click NBRGB button

Step 6: When you find one that makes you happy, click magic OK button

Step 7: You now have a new image called NBRGBCombination

I snuck inside for a minute to try these settings.

 

Amazing! and so easy following the steps that you so kindly laid out. Of course, by linear image is not balanced yet, but the difference (with default bandwith) is very apparent!.

 

Right image is osc. Left image is osc with the L-eNhance R channel added. 

 

NBRGBCombination.jpg

 

Thanks a bunch Johnny! I appreciate the help!

 

Jim


Edited by Juno18, 12 May 2021 - 03:40 PM.

  • jonnybravo0311 likes this

#9 Juno18

Juno18

    Mariner 2

  • *****
  • topic starter
  • Posts: 287
  • Joined: 07 Dec 2018
  • Loc: Long Beach, Mississippi, USA

Posted 12 May 2021 - 03:40 PM

This is the way I like to do this, it is an iterative process to taste:

Split all images into R, G, and B, making sure they are all registered to one another.

In PixelMath place in the R, G, B images from the OSC.

Starting with the NB R image combine it with the OSC R image in two different ways, [1] weighted average, [2] max pixel value, here is an example of each:

[1] R*0.6 + Ha*0.4

[2] max(R, Ha), you can also weight within the max function: max(R, 1.2*Ha)

Compare the new image to the the base line OSC image and look for differences that you like, you can do this by toggling one image over the other,

To toggle images place one image on top of the other, then hit <Ctrl><PgDn> and it will blink the images to see the effects.

Once you find a new image that you like, this becomes the new base line image.

Now rinse and repeat for the G channel using whatever NB images you choose, then for the B channel.

 

Takes time, but if you are patient it will yield very good results.

Much appreciated Peregrinatum! I will try this method out tonight or tomorrow!

 

Thanks a bunch for replying and your very helpful method.

 

Jim




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