Deconvolution drains color from image

[astrovienna]

I've been messing around a bit with my narrowband M1 image, following the suggestions in Mike Wiles' tutorial.  I'm deconvolving the unstretched, color combined image, using a PSF generated from about 50 stars in the field, with local deringing support using a star mask.  The deconvolution gives nice results, except that some of the color disappears from both the nebula and the stars.  This is particularly noticeable in green and yellow.  Is that normal?  If it helps to diagnose this, the process console notes "local divergence" during 2-3 iterations in channel 2.  If instead I do the deconvolution post-stretching, using a parametric PSF, I don't notice this color loss.

Kevin

[RickS]

Did you specify Luminance as the target for the decon?  If you're applying it to the RGB/K components then I wouldn't be surprised if there was some impact on colour.

Cheers,
Rick.

[astrovienna]

Thanks Rick!  That worked to preserve the colors, although I did have to drop the global dark setting in deringing by a factor of 20.  I also had to rebuild the star mask for local deringing support to be much more aggressive, and I could use some help there.  I can't seem to make the mask aggressive enough to protect some of the brightest stars.  I'm making the mask with an MLT starlet transform with 8 layers, first and residual deselected, and then a star mask following the suggestions in the M81/M82 walk-through.  Any suggestions for protecting the biggest stars?

Kevin

[RickS]

Glad I could help with the decon...  You can try stretching your star mask with a HistogramTransformation to strengthen it.  MorphologicalTransformation can also be helpful to adjust a star mask, e.g. Dilation to increase the size of the protected areas.

Cheers,
Rick.

[astrovienna]

Thanks again, Rick.  That helped.  One last question:  in the future, am I better off deconvolving each channel separately?

Kevin

[RickS]

Thanks again, Rick.  That helped.  One last question:  in the future, am I better off deconvolving each channel separately?

I have occasionally used decon on individual channels to match star sizes but most of the time I just apply it to the luminance (I typically process luminance and colour separately.)  I'm sure that either approach can work.

Cheers,
Rick.