Normalizing mosaic ...
 
Share:
Notifications
Clear all

Normalizing mosaic from color camera with RGB images and Ha images


(@h-dominik)
Hydrogen Atom Customer
Joined: 4 months ago
Posts: 3
Topic starter  

Hello!

I am pretty new to APP and I have a question about processing a mosaic with was taken with a color camera (Nikon D810A). I have two image sets, one with normal color RGB data and one with filtered Ha data.
I have loaded all images and registered all at once to create then two mosaic images which I later want to combine into one final image.

Unfortunately when it comes to normalize the backgrounds (step 5) which I really want to do to improve the result and make stretching easier I cannot define a reference image separately for each of the image sets. If a Ha image is set as reference image then all RGB images turn into black/red and if an RGB image is set as the reference image then the Ha images turn into weird colors. Maybe this may be a not too big issue as Ha is one single channel anyway, but it confuses me a bit when I see the preview.

When I only normalize all Ha images first and then change the selection to the RGB images and assign the reference image to an RGB image then all normalization data get deleted.. so no luck here.

How can I solve this?

Thank you and regards,
Dominik


ReplyQuote
Topic Tags
(@vincent-mod)
Universe Admin
Joined: 5 years ago
Posts: 5130
 

I would first create the mosaics separately, saving them as RGB and Ha. Then load those together and register them to each other and save the registered files (in the register tab, all the way down). These saved files can then be loaded in the combineRGB tool, when you load the RGB it will pop up 3x to load the R, G and B channels and then you load the Ha. Then you select a preset you like.

You can also split the RGB mosaic in R, G and B channels (tab 2, all the way down) and then save the files, you can then load these in again as mono data, together with the Ha result and then normalize them against each other (tab 5, save normalized frames). Then combining those in the Combine tool.


ReplyQuote
(@h-dominik)
Hydrogen Atom Customer
Joined: 4 months ago
Posts: 3
Topic starter  

Thank you for the reply!

I tried to create the mosaics separately, but then they wont' register to each other. The mosaic is a 3x3 wide field with 85mm, probably the distortion between the mosaics is too big.

I then tried to split the Ha data into it's channels as you recommended and selected the RGB images and the Ha red channels only, but then I needed to restart APP as even when the RGB images and the Ha red channel images were correctly selected in the list APP ignored the Ha red images in star analysis and registration.

After restarting and adding the RGB images (NEF files) and Ha red channel images (FITS files from the channel split) the star analysis and registration succeeds for all images but then normalization fails with the error message

Encountered error in module:
NormalizationActionListener

Error message:
java.lang.ArrayIndexOutOfBoundsException: 1

Cause:
1
E.a(Unknown Source)
E.doInBackground(Unknown Source)
javax.swing.SwingWorker$1.call(SwingWorker.java:295)
java.util.concurrent.FutureTask.run(FutureTask.java:266)

I tried to restart APP and doing exactly the same, which also failed with normalization, but this time with a different stack trace:

Encountered error in module:
OverlapBetweenTwoImageObjectsCreatorWorker

Error message:
java.lang.ArrayIndexOutOfBoundsException: 1

Cause:
1
java.awt.image.ComponentSampleModel.getSampleDouble(ComponentSampleModel.java:872)
java.awt.image.Raster.getSampleDouble(Raster.java:1690)
G.a(Unknown Source)
G.doInBackground(Unknown Source)

 

and just retry normalize with out restarting APP then:

Encountered error in module:
OverlapBetweenTwoImageObjectsCreatorWorker

Error message:
java.lang.ArrayIndexOutOfBoundsException

Cause:
null
no trace

 

Log output in the main window (I just tested with two mosaic positions to see if it works):

 

2347 - 5) NORMALIZE: starting...
2347 - 5) NORMALIZE: reference frame found: /Volumes/SSD-HD-2/AstroPixelProcessor/Test2/RGB1__D8A3785.NEF
2347 - 5) NORMALIZE: advanced normalization mode: analyse reference
2347 - 5) NORMALIZE: entering advanced normalization mode.../Volumes/SSD-HD-2/AstroPixelProcessor/Test2/RGB1__D8A3785.NEF
2347 - 5) NORMALIZE: loading the reference frame : /Volumes/SSD-HD-2/AstroPixelProcessor/Test2/RGB1__D8A3785.NEF
2352 - 5) NORMALIZE: converting the reference frame to 32bits normalized floats...
2352 - 5) NORMALIZE: analyzing the reference frame...
2355 - 5) NORMALIZE: file: RGB1__D8A3785.NEF : band 1 reference location and scale : 1,0071E-01 - 1,8550E-02
2355 - 5) NORMALIZE: file: RGB1__D8A3785.NEF : band 1 this frame location and scale: 1,0071E-01 - 1,8550E-02
2355 - 5) NORMALIZE: file: RGB1__D8A3785.NEF : band 1 MRS gaussian noise 3,2057E-03 percentage of pixels 11,046 %, scales used 4
2355 - 5) NORMALIZE: file: RGB1__D8A3785.NEF : band 1 SNR 1,0711E+015) NORMALIZE: file: RGB1__D8A3785.NEF : band 2 reference location and scale : 1,5027E-01 - 2,8782E-02
2355 - 5) NORMALIZE: file: RGB1__D8A3785.NEF : band 2 this frame location and scale: 1,5027E-01 - 2,8782E-02
2355 - 5) NORMALIZE: file: RGB1__D8A3785.NEF : band 2 MRS gaussian noise 3,6541E-03 percentage of pixels 18,987 %, scales used 3
2355 - 5) NORMALIZE: file: RGB1__D8A3785.NEF : band 2 SNR 1,0833E+015) NORMALIZE: file: RGB1__D8A3785.NEF : band 3 reference location and scale : 1,0755E-01 - 1,6357E-02
2355 - 5) NORMALIZE: file: RGB1__D8A3785.NEF : band 3 this frame location and scale: 1,0755E-01 - 1,6357E-02
2355 - 5) NORMALIZE: file: RGB1__D8A3785.NEF : band 3 MRS gaussian noise 3,9189E-03 percentage of pixels 15,478 %, scales used 4
2355 - 5) NORMALIZE: file: RGB1__D8A3785.NEF : band 3 SNR 8,2577E+00
2355 - 5) NORMALIZE: advanced normalization mode: normalizing to reference frame
2302 - 5) NORMALIZE: file: RGB2__D8A3786.NEF : band 1 reference location and scale : 9,5956E-02 - 1,7331E-02
2302 - 5) NORMALIZE: file: RGB2__D8A3786.NEF : band 1 this frame location and scale: 9,6101E-02 - 2,2243E-02
2302 - 5) NORMALIZE: file: RGB2__D8A3786.NEF : band 1 MRS gaussian noise 2,5888E-03 percentage of pixels 10,400 %, scales used 4
2302 - 5) NORMALIZE: file: RGB2__D8A3786.NEF : band 1 SNR 1,1716E+015) NORMALIZE: file: RGB2__D8A3786.NEF : band 2 reference location and scale : 1,3710E-01 - 2,9058E-02
2302 - 5) NORMALIZE: file: RGB2__D8A3786.NEF : band 2 this frame location and scale: 1,3919E-01 - 3,3259E-02
2302 - 5) NORMALIZE: file: RGB2__D8A3786.NEF : band 2 MRS gaussian noise 3,2101E-03 percentage of pixels 14,936 %, scales used 4
2302 - 5) NORMALIZE: file: RGB2__D8A3786.NEF : band 2 SNR 1,1593E+015) NORMALIZE: file: RGB2__D8A3786.NEF : band 3 reference location and scale : 9,8146E-02 - 1,6366E-02
2302 - 5) NORMALIZE: file: RGB2__D8A3786.NEF : band 3 this frame location and scale: 9,9229E-02 - 1,5212E-02
2302 - 5) NORMALIZE: file: RGB2__D8A3786.NEF : band 3 MRS gaussian noise 4,2524E-03 percentage of pixels 12,491 %, scales used 4
2302 - 5) NORMALIZE: file: RGB2__D8A3786.NEF : band 3 SNR 8,5281E+00

 

I have not yet tried to split up the RGB images and do it then like described in https://www.astropixelprocessor.com/how-to-create-a-multi-channel-mosaic-by-mabula/, but I fear that this would then consume a lot more calculation power as the mosaic parts are then artificially a lot more.

Using the original Ha raw files instead of the split up red channel does not cause the error in normalization and the mosaics are perfectly aligned.

I tried another approach:

1.) Loading RGB images, analyze stars, register, normalize and save normalized without register info (I thought without as the mosaic position will be re-calculated later with all files, or is with reg info better for more accurate normalization result?)
2.) Separately do the same for Ha images
3.) Use the normalized files and register them again, but this time all together (needed to restart APP before this, as registration said no files to register despite the normalized files were selected, maybe a bug?)
4.) Now disable normalization and integrate the two mosaics after each other

Seems a bit complicated, but with this I could get the desired result (fine normalized seams in all parts and no "discoloring" of the Ha mosaic).

Is there a possibility to get this process a bit easier to do without the use of "workarounds" in a future version of APP?

Bwt. I really like the seamless stitching capabilities of APP.

Thank you

This post was modified 3 months ago 2 times by Dominik Hölzl

ReplyQuote
(@vincent-mod)
Universe Admin
Joined: 5 years ago
Posts: 5130
 

What I meant was to split the RGB image into it's channels. These will then become mono channels and can be normalized together with the Ha mono channel.

But yes, we do have it on our todo list to make this an easier workflow.


ReplyQuote
Share: