Until now I stack and process my L-extreme data like a normal OSC file and use the OSC_preprocessing script. When done I extract the rgb channels and recombine as SHO. Is there any benefit by using the OSC_Extract_HaOIII script?

Yes, an important step is not done in this script: demosaicing, also called debayering. This creates 3 monochrome images from incomplete filter data, so many pixels have to be invented based on the neighbours. Extracting the two filters doesn’t need that, but doesn’t give stars as round because one pixel is missing for OIII, the red one.

That makes a lot of sense. I will try this script and see what I can get.