Loss of colour in PixInsight processing

[cedric_r]

Hi all,

I'm kind of stuck with a picture: when i process it, I lose a lot of the colour and I can't figure out where and why. Here is the result of integration, DBE, PCC (as Sb galaxy, but average has the same result), deconvolution, some noise removal on the backround, and stretching:


As you can see, M63 has little colour and i find it impossible to extract it. Can someone help? My processing normally goes ok, but not this time. Any ideas what would result into this? Or what I'm not doing that I should?

For reference, the original integration file is here (not the drizzle integration one, it's 1GB!). it has a strong red cast (don't know where it comes from, it wasn't apparent in the original images), but it goes away pretty well with a single DBE and shouldn't matter too much. How would you go about it?

For info, my pre-processing was imageCalibration, CosmeticCorrection, and Debayer for each individual nights with their bias,darks, flats; then SubframeSelection, StarAlignment, LocalNormalisation, ImageIntegration, and finally DrizzleIntegration.

Thanks for your input guys.

[pfile]

the color is there - however there seem to be some problems with the data - strong, repeating mottling in the background in all colors and yes despite color calibration there does seem to be a strong red cast. in addition the core of the galaxy is overexposed.

i had to repeatedly apply saturation curves and boost the blue while pulling down the red to get into the neighborhood of how M63 should look...

rob

[cedric_r]

Thanks Rob. That's much better. I've noticed the mottling on the background. I have no idea where that comes from. Do you know what could be the cause?

I assume the core is blown because I was using 10 minute exposures. I should make shorter exposures and merge.

[pfile]

i guess as a start i would check if you have any subexposures with uneven illumination from clouds or dew...

rob

[cedric_r]

Thanks. I'll check for that.

[sharkmelley]

I agree that it looks mottled - that needs a bit of investigation. 

I very quickly processed it as follows:
1) DynamicCrop to remove the messy borders
2) ColorCalibration using all the defaults.  It scaled the RGB channels as follows: 1.0, 1.5, 2.8
3) AutomaticBackgroundExtractor with function degree set to 1
4) HistogramTransformation setting the Highlights to 0.5  (this prevents the star centres and galaxy centre being discoloured.
5) 2 iterations of ArcsinhStretch - both times using "Estimate Black Point" button in the preview and both times setting Stretch factor to 7

The background is still a bit blue/purple.  That could be fixed by using a more careful DBE in place of the quick and dirty ABE.

Mark

[cedric_r]

Thanks very much Mark. I've never looked at ArcsinhStretch and I also like where your version is going. That's my weekend busy

[cedric_r]

Hi guys,

Here is the first attempt at reprocessing adding LinearFit and ArcsinhStretch into my process:


The colours are definitively back and I'm happier with the results. There is still a lot that needs to be done to make it a better picture (e.g. I clearly have saturated stars that end up with weird core colours), but my problem is solved.

Thanks again, that was very interesting and I got a lot of useful information out of your comments.

[sharkmelley]

Wow that's a whole lot better.  You definitely have colour!

The weird colours in the star cores arise from colour balancing saturated data.  If the star core is saturated on the sensor then it will have RGB value of (1.0,1.0,1.0) on a 0.0-1.0 scale.  The colour balancing applied multipliers of 1.0, 1.5, 2.8 so it ends up as a greenish blue.  I simply truncate the upper values of the data to make the cores white again - that was the purpose of setting Highlights to 0.5 with HistogramTransformation.

There are probably other treatments you can apply instead, once you understand the cause of the weird cores colours.

Mark