RGB channels in star cores, purple cores

[AstroScience]

Hi guys,
when I check single frame for that specific star shown on image, I can see that all 3 channels are overexposed. But after stacking, Background Neutralization and Color Calibration, RGB channels shifts in cores causing them to be purple. Shown on image is linear image without STF.
If I stretch manually with HTT, then RGB values got 0.99, 0.97 and 0.98 and became white, but if I using Masked Stretch it leaves me with stretched image but purple cores of stars.  How can I fix this?

[Carlos Milovic]

First of all, information is lost. From that point, we may go forward using two approaches:
- Move the white point to the left (in the HistogramTransform) so all saturated data is pure white. You may end loosing some details in the star cores, clipping more data, but you won't have any cast.
- Be more conservative about that, and don't care too much about star cores. Then, use the script that fixes the hue of saturated data (can't remember the name right now), or use a mask and try to do that manually.

Of course, the best course of action would be to always take shorter exposures, so an HDR composition will allow you to avoid any saturated data.

[Phil Leigh]

Is that the Repaired HSV Separation script? - I was hoping that script could repair the pink cores but I'm not having much luck with it... but that's probably down to my data or I'm just doing it wrong? I still end up with the star centre colour problem.

[Bob Andersson]

Hi Phil,

The Repaired HSV Separation Script might be able to help but it can't work miracles and I've found that it really helps to get the white balance as good as you can before using it. I find that even after using ImageCalibration it helps to have a really close look at the histogram of the image and make any adjustments as necessary to get the RGB curves aligned as appropriate for the subject in the image. All this on the linear data, of course. Once the white balance, and any other manipulations that need to be done before stretching, is done then try the Repaired HSV Separation Script. Experimentation with the settings is mandatory, I'm afraid, but you might be surprised how aggressive you can be with the Repair Level and Max Repair Radius sliders - the readout under the Repair Level slider for the number/percentage of pixels likely to be changed gives a good clue to an initial value. But even in best case scenarios I find that there are usually a number of stars that need fixing manually.

Bob.

[Phil Leigh]

Thanks Bob - I will persevere with the script using your tips.. thanks very much,
Phil

[AstroScience]

Thank you guys for the suggestions, after trying the script I could fix most of the saturated stars, but some of them, including the Supernova seen in the image, get artifacts. I guess I need more time with the script. Meantime I can get very good results by creating the star mask for all those saturated stars, while in linear stage and adjust manually RGB channels with HT so the saturated areas have even values, then use MaskedStretch and the results not bad at all.
Besides that, I can stretch the image the regular way, without using MaskedStretch and then push back the blown stars with MT. So many ways to deal with the problem, can't express my love for PI.