Each one give his own interpretation in the image elaboration with narrowband, sometimes going out of the "classic" combination. I like to think I'm not doing scientific report with my image. I simply can't.
In answering to the question, what bring out details is usually the luminance.
So, once combined and optimized the NB image, it can be extract the luminance from.
This luminance, if needed with a small "cleaning", is then used to emphasize the NB image adding the (!) Luminance part.
Another way with the H and O files is to combine them in a mix that depend on what you have imaged.
Adding the pure H data as luminance it can not show all the details, especially if the subject is rich in O3. So a mix of them in a percentage that can vary, is a good choice.
I feel comfortable using the SHO script of Pixinsight (can be found in PI forum). It allows more control.
And last, I will never stop to thanks Luca, an Italian guy living in US for teaching me what above.
Another small note is the noise reduction that I use. You can search for the "Jon rista method". Jin is a great photographer and experienced in astrophoto.