Incorrect RGB Colours after PixInsight Channel Combination - Forums

Topic starter

Jul 31, 2023

Hi All,

I'll try to keep this short; I'm relatively new to mono imaging and I think I am doing the correct PixInsight workflow for channel combination as follows:

1/ WBPP of the R, G and B channel data (in this particular case it's M57 with 20 x 300 seconds on each channel at 100 gain on the ASI6200) - I'm using the same gain for bias, dark and flat frames; (dark is 300 seconds too; all at minus 10 C)
2/ Star Alignment of the master channels
3/ Channel Combination (I've tried with and without Linear Fit)
4/ SpectroPhotometricColourCombination / Background Neutralization

ABE / BlurX / NoiseX etc have no effect on any of the above, before during or after; oh and all of the above is in linear mode

Bottom line is that the colours are all wrong; the centre of M57 is an ugly purple and the outer fringes are a greenish / yellow - no blue or red anywhere; very strange. Curves, Saturation etc. can't seem to fix it either.

Any suggestions would be appreciated; I'm sure it's something simple that I've somehow missed along the way?

Thanks

Pete
PS: The green channel histogram was quite a bit to the left of the red and blue even at the same time and gain but I've seen other examples here on AStrobin where all channels are the same with good results (maybe some balance issue?)

Helpful Engaging

andrea tasselli

Frequent contributor

Jul 31, 2023

Most likely you got the channels mixed up. But without an actual image to look at that's difficult to judge.

Well Written

Wim van Berlo

Jul 31, 2023

As Andrea wrote, probably mixed masters in channel combination. You can examine the header information of your subs to see if they are ok. Load a sub into PixInsight and go to Files > Fits Header. There you should see which filter was used. Worst case, your filter identifiers are incorrect. This happened to me once after I upgraded Kstars/Ekos, and the filter labels in the software didn't match the physical filters in the wheel. The FILTER keyword in Fits Header won't be correct in that case.
To fix your current problem, try different combinations of channel combination. The colours that you do have in your current image, should give you an indication of how to swap. After doing background neutralisation, the colours should be nearly correct. The core of the brightest stars may be off if you have light pollution.

cs,
Wim

Helpful

Torben van Hees

Jul 31, 2023

You do a new, channel-linked screen stretch after the SPCC, right?

Vince

Jul 31, 2023

You could watch some of the videos from Adams Block on Youtube.
They are a bit verbose but the content is second to none. You will need to check which video is the one you require.
(1) Adam Block - YouTube

Concise

Otto Vehviläinen

Jul 31, 2023

Not sure how much light pollution you have, but I’d recommend doing background extraction before color calibration as it can have a huge effect on the end result especially if there’s a lot.

Helpful Insightful Respectful

sletmoep

Jul 31, 2023

I think your P.S. comment holds the secret. Check out this discussion related to PCC and purple images:

https://pixinsight.com/forum/index.php?threads/photometric-color-calibration-resulting-in-pink-purple-stars.16514/page-2

Etienne MARQUET

Jul 31, 2023

=MsoNormalBonjour,=MsoNormalIl faut tout traiter en non-linéaire. Ajuster au mieux chaque couche (L, R, G, et B) pour avoir de belles couches. La L est la + importante.=MsoNormalEnsuite, combiner avec LRGBConbination, et ajuster les curseurs de chaque couche.=MsoNormalÇa devrait donner de meilleurs résultats.=MsoNormalHello Everything must be treated in non-linear. Adjust each layer (L, R, G, and B) to have beautiful layers. The L is the important +. Then, combine with LRGBConbination, and adjust the sliders of each layer. It should give better results.

Peter Hannah

Jul 31, 2023

I agree it's worth checking your filter labels match the physical filters. You don't mention which acquisition software you used but in Maxim DL for example they are defined in Camera Control/Set up Filter.

My filters were once put back in the wrong slots after cleaning at my remote observatory. Took me ages to figure out what was wrong. Then I had to edit all the FITS headers of the affected subs…

Well Written Helpful Insightful Concise Engaging Supportive

Dave & Telescope

Jul 31, 2023 · Edited: Jul 31, 2023

Agree with all of the above. Couple more things.

1) Try looking at the color right after channel combination before anything else. And yes you are correct you do not need linear fit for that step. See my attached screenshot. First unlink the channels (top black arrow), then restretch with the radioactive icon (lower blue arrow). If you see the colors look reasonable, then
you're probably doing something wrong in the later steps. If the colors are still way way off then check some of those other areas, filter designations, headers etc

2) Try using basic Color Calibration, not SPCC and definitely not PCC for the reasons already mentioned

HTH
Regards
Dave

InkedSTF.jpg

Helpful Supportive

AstroRBA

Topic starter

Jul 31, 2023

Thanks for all of the info:

I shoud note that the basic colour calibration and neutralization and various workflow positions of ABE etc., also do not yield the proper results and the FITS headers all re-confirm correct colour filter selection AND I double checked EFW slot positions as well; but Just for trial and error, I also tried all other colour combinations in channel combination and none resulted in a correct appearance.

Even before applying any processing it's clear that something is skewed. Just for trial and error, I also tried all other colour combinations and none resulted in the correct appearance. No matter what I do I can't get rid of that ugly central colour.

Evan though I'm sure it's still something simple, I've tried to eliminate the simple things (I think?!)

Sean Mc

Jul 31, 2023

Going to go out on a limb here… did you blink through your lights?

François Guiot

Jul 31, 2023

Hi,

As suggest by Dave:

Dave & Telescope:
InkedSTF.jpg

What happen when you do ChannelCombination (no processing at all) and then, simply use ScreenTransfert Function (unlinked, so each channel is auto stretch with its own function) ?

If the colors are correct => It's something about SPCC; probably with the white reference (should be average spiral galaxy).
If not well... Something goes wrong during preprocessing

Regards;

AstroRBA

Topic starter

Jul 31, 2023

Hi,

As suggest by Dave:
Dave & Telescope:
InkedSTF.jpg

What happen when you do ChannelCombination (no processing at all) and then, simply use ScreenTransfert Function (unlinked, so each channel is auto stretch with its own function) ?

If the colors are correct => It's something about SPCC; probably with the white reference (should be average spiral galaxy).
If not well... Something goes wrong during preprocessing

Regards;

Tried that - no difference

Wim van Berlo

Jul 31, 2023

Since we have eliminated the most obvious, I think we need what @andrea tasselli wrote earlier: show the image in this thread. Without it, I don't think we can help you further.

cs,

Wim

Well Written Concise

Tom Engwall

Aug 1, 2023

Do not apply an STF after Noise X. Just do a HT to make the image non linear before doing any more processing.

AstroRBA

Topic starter

Aug 1, 2023 · Edited: Aug 2, 2023

Update:

Thanks everyone for all of the info; since my original post I've collected quite a bit more data over the last few nights and I'm going to start from scratch begining with a whole new WBPP approach to see if some early step may have gone wrong - I've now got H and O data to add for more fun and games!
I'll keep all of the above in mind !

Pete

Tim Hawkes

Aug 1, 2023

Sometimes the error can be as far back as data acquisition? I have once - to my shame - collected two 2 sets of blues (thinking I had swapped the blue filter out for green and processing the second acquisition as a green). Fortunately only flats but it can happen

Tim

midnightsnacks

Aug 4, 2023

Try ABE before SPCC.

AstroRBA

Topic starter

Sep 7, 2023

Update:

I shelved all of the M57 data for now (I'll return to it some day!) and moved on to M27 - I lowered the exposure and gain (from 300 / 100 to 180 / 50) and also collected some H and O data (L too using the LPro to kill some of my brutal B8 LP) - I think the earlier M57 data was just way too saturated at gain 100.

Although still difficult, I managed some better colour balance on M27 (but still have work to do on the star colours) - It seems that a lot of experimenting is neccessary to land the right data set for this rig! Still much better than watching crap on TV!

Thanks again everyone for the suggestions!

Pete

Helpful Respectful Engaging

Bob Lockwood

Sep 7, 2023 · Edited: Sep 7, 2023

I'm going out on a limb hear as I don't know if this affects anything. You are using an L-Pro instead of a lunminance filter.
Will this not affect the image while making a color image, since there basiclly two different filters and blocking different light.

AstroRBA

Topic starter

Sep 7, 2023

Bob Lockwood:
I'm going out on a limb hear as I don't know if this affects anything. You are using an L-Pro instead of a lunminance filter.
Will this not affect the image while making a color image, since there basiclly two different filters and blocking different light.

Hi Bob, I did try processing with and without thre LPro "Luminance" data; although it mostly helped the M27 data set overall, I'm currently working on the Cocoon Nebula and I find that the LPro it doesn't pick up much at all in this case? Even at 900 seconds gain 200; a bit odd but I think that I need some LP filtering with this brutal B8 zone that I'm in. I might skip L on this Cocoon attempt and see how H and O collect?
Thanks,
Pete

Jerry Yesavage

Sep 7, 2023

Tim Hawkes:
Sometimes the error can be as far back as data acquisition? I have once - to my shame - collected two 2 sets of blues (thinking I had swapped the blue filter out for green and processing the second acquisition as a green). Fortunately only flats but it can happen

Tim

Right filters in right slots? Chased my tail on one of these once.