Problem with blue star edges

[wkrispler]

Hello,

I would need your help concerning some artefacts on stars.
In the attached image (3:1) you can see some stars with blue edges on the left top side and some stars with yellow/brown edges on the right bottom side.
It is still in the linear phase.

The following steps have been applied
*Color Channel Alignment (Extract, Align, Combine) as I thought this is a good measure to prevent such artefacts
*Background neutralization
*Color Calibration (the histogramm of the image looks perfect, the color channels are 100% aligned)
*Color Saturation of the image AND the stars with curves transformation

I played around with Channel Match but that did not help.

Any idea where this comes from and what can I do?

Thank you,

Wolfgang

[aworonow]

Was this shot with a one-shot-color or DSLR camera? If so, and if the images span a range of altitudes above the horizon (including moderate to low altitudes), then such artifacts are inevitable, and arise because the amount of atmospheric refraction distorts images higher in the sky less than it does those lower in the sky. I would still think that taking the color channels apart and treating them separately would remediate this problem--but maybe not ??? Have you tried using the components of the same frame as the reference frame for each R,G,B alignment? Maybe the attached, which I put together as part of a talk, will help you visualize the issue and track down the problem.

Finally, unfortunately, it could be some lens problem...hope it's not!

Alex W.

[wkrispler]

Hi Alex,

thank you for your feedback.
The image was taken with a DSLR starting at an altitude of 66° going down until 25°. This may be the cause of the problem as explained by you!
The optics is ok , I had it checked recently!

You mean I should seperate the channels BEFORE the image integration. I , after calibration, should split every single light frame in the color channels, align them and combine them again and then do the image integration?

Thank you,

Wolfgang

[aworonow]

Wolfgang,

This is the procedure I would follow:
The sequence could vary depending on whether you shot raw frames (still with the Bayer pattern) or if you have frames the camera deBayered for you (probably jpg). The former is preferred.
But for now, just go through your normal preprocessing calibration routine that ends with a bunch of calibrated color images--no alignment yet!. Then generate separate RGB images for every frame (there is a BatchChannelExtraction script).
align the R, G, and B channels separately, and make separate integrated (stacked) R, G, and B images. Then align those three and use the LRGB or other channel combine procedure, to get the color image. That should help, I hope.
Good Luck, Alex W.

[msmythers]

Here is one thing I do that can help at times. Before any post-processing extract the 3 channels from the stacked integrated image. Open StarAlignment. Use the green channel as the reference. Select Distortion correction. Now Align the red and the blue channels to the green. Recombine with the ChannelCombination. You might need to adjust the parameters for the Distortion correction. You might also try a different Pixel Interpolation in the Interpolation section of StarAlignment.



Mike

[wkrispler]

Thank you all very much for your reply. I will try all suggestions!

All the Best,

Wolfgang

[RickS]

If, after realignment, you still have (hopefully much smaller) colour fringes another useful technique is to apply a star mask then use one of the wavelet processes, e.g. MultiscaleMedianTransform, and remove the bottom few layers with the Target set to Chrominance.  This will blur the colour in your stars and reduce the fringes further.  You may want to bump up the saturation of the stars with CurvesTransform afterwards.

Cheers,
Rick.