PixInsight equivalent to Lightroom's edge removal

[ruediger]

Hi all,

I like to do widefield shots with 24mm to 135mm lenses and sometimes end up with colored edges around the stars, due to imperfect focussing or imperfect lenses. In Lightroom, there is an edge removal function in the lens correction section that works perfectly, see attached screenshot [shown is a 2:1 detail of a 24mm milky way shot].

I tried to do this kind of edge removal in PI, but failed so far.

Anyone has some suggestions?

Thanks in advance,
  Rüdiger

[Alejandro Tombolini]

Hola Rüdiger,

Quizas con una máscara de estrellas protegiendo el fondo y moficando S y c en las curvas.
Tendrías que trabajar mas que nada sobre la máscara; en el ejemplo está con los valores por defecto y solo afecta a las estrellas más grandes.
También podrías trabajar con la herramienta "MorfologicalTransformation" sobre las estrellas manteniendo la misma máscara. 


Hi Rüdiger,

May be with a star mask protecting the background y then modifying S and c in Curves.
You would have to work mainly in the mask; in the example is with default values and only afect to big stars.
Also you might work with the tool "MorfologicalTransformation" on the stars supporting the same mask.


Saludos.
Alejandro. 

[zvrastil]

Hi.

I believe there's no dedicated tool for this task in PixIsnight. Thinking about it, I have some ideas how to improve this problem "manually". I'm not sure which kind of problem you're fighting for. I've seen (in my images taken with older lens) three possible types of this problem:

1) Stars far from image center have two-color edges - edge closer to the center is red while the opposite edge is blue (or vice-versa) - this means lens are not well corrected for chromatic aberration further from the optical axis. This is very common even for today low and mid range lens. I'd try to fix it by splitting image into RGB channels, aligning Red and Blue channel to Green channel using StarAlignment and combining them back into RGB image.

2) Most stars have edge of same color (typically blue). This is again result of uncorrected chromatic aberration but this time in whole field of view. This often happens with older lens (made for film photography). My guess is it could be improved again by splitting RGB image into channels and reducing star sizes in "incorrect" channel (like blue one) with either Deconvolution (hard way) or Morphological transformation (easier way).

3) Stars in one part of the image then to have edge of one color (like blue), while stars in other part of the image has edge of other color (like red). Lens have problem #2 but also focal plane is not perpendicular to image sensor. This again happens with older or low-range lens due to mechanical imperfections. I have no idea how to fix this problem correctly, other than some clever masking to cover only star edges and decreasing saturation using this mask. Such mask could be constructed with StarMask, MorphologicalTransformation and PixelMath.

regards, Zbynek

[ruediger]

Thanks for your suggestions, I will try out and experiment in the next days.

Since I organize all my pictures (daylight and astrophotography) in Lightroom, I could do correct lateral chromatic aberrations and remove colored halo around stars easily at the very last processing step. But I guess, these corrections should be done as early as possible, i.e. directly after stacking, and operations like deconvolution or morphological operations would perform better on corrected data.

Regards,
 Rüdiger