Green Cast on PI SPCC with APP stacks

Just finished stacking with new Beta 17 and did SPCC in PI - no, it does not work, sorry.....

How can this be resolved?

Chris

I have no experience with the SPCC tool, that may better be asked on the PixInsight forum. But a green cast is normal as you have twice as much green pixels in the bayer matrix. If you perform normalization in APP, this should be taken care of. Can you share screenshots as to how your images look like and tell me more about the settings used in APP?

Vincent, thanks very much for your quick reply!

The issue is not the green cast after stacking, on linear images of OSC Bayer-matrix cameras - I am aware of this.

The issue is the difference in outcome when performing SPCC on OSC-images stacked with APP or stacked with PI.

Theoretically, the outcome should be identical, but on images stacked with APP there is always a green cast.

There is no problem with the colors of the integration result in APP, I actually like the result (after light pollution removal = background neutralisation). But if I enter this image into PI and run SPCC on it, a result with greenish cast comes out. Same thing when using PCC instead of SPCC.

Again, this is only OSC-related - no difference in outcome with mono-data.

The manual for SPCC explicitly says:
https://pixinsight.com/doc/docs/SPCC/SPCC.html

"Another important point when working with OSC cameras is to always use drizzle. Besides the fact that only a drizzle integration can provide optimal results by avoiding interpolation, some de-Bayering algorithms may modify color proportions at small scales, where interpolation of missing color data in CFA patterns plays an important role. This is the case with VNG. While large–scale structures remain untouched, point sources (like stars) will be altered. And, what is worse, this color shift depends on the sizes of the stars so that each image can be altered differently. This affects stellar photometry so that de-Bayering can lead to unexpected erroneous results, both with PCC and SPCC. You can see two examples below; the first one is with the same Messier 33 image used previously"

And it shows two example pictures of M33 and M31 with drizzle and VNG-de-bayering.

So I tried a drizzle integration in APP with unfortuantely no other result. So I figured it must have something to do with the de-bayering in APP.

I know that many astrophotographers like stacking in APP because it is much faster and much more convenient than in Pixinsight (especially multiple sessions).

After that, they switch over to PI for further processing (for tools that are only available in PI, like BlurXterminator, for example).

But if the OSC APP-result cannot be color calibrated in PI propperly, I will no longer have any use for APP, unfortunately.

Maybe somebody has found a solution to this?

Thanks & CS!

Chris

I've informed Mabula to shine his light on this as well. 😉

Hi @chrisl,

Thank you very much for reporting this.

Yes, performing demosaicing in OSC data has the mentioned affect. It affect correct color in structures that have sizes of 1-5 pixels more or less I would think.

The key here is that drizzle in APP still performs debayering/demosaicing and then drizzles the debayered results. You need to select Bayer/X-Trans drizzle for demosaicing not to occur. Did you try this as well? Perhaps  you even need to do is with larger droplets than the default of 1 pixel. I would suggest to try Bayer/X-Trans drizzle with droplets of 2 and then perfrom the star color calibration with the PI tools and let us know if the problem is still there or not. If the problem would still be there, it would be odd since no demosaicing would have been performed.

Mabula

Mabula,

thanks very much for your quick reply!

I just made a new integration using Bayer/Xtrans drizzle with the parameters you suggested - unfortunately it did not work out.
I used the latest release Beta 17, BTW.

To clarify what I mean with "green cast" I attached a picture comparing 3 images (same subframes):

1.) Stack "as is" out of APP using Bayer XTrans drizzle integartion

2.) The above color calibrated using PI SPCC

3.) Same data but stacked in PI and color calibrated with SPCC

Remark on 3.): I did not use flat frames in this integration, because I only kept the APP master flat and not the individual flat frames. Unfortunately this master flat from APP was made in 16 bit depth and not in 32 bit, so it was not usable in PI. Master calibration frames from APP are usable in PI, by the way, as long as they were compiled in 32 bit. But this is a different topic.

What is astounding is, that there is a difference in color at all. Very strange....

In case you would like to have a look at the fits files of these 3 pictures, I have uploaded those on my Dropbox:

https://www.dropbox.com/s/nfz4lt4p4d6mn1a/FITS%20files.zip?dl=0

Thanks & best regards!

Chris