Built-in collision detection - what are the limitations?

Hello all! I am new to CF 2.5 so please forgive me if these are newbie questions. X)

I played a fair bit with the free version before buying and have looked at a lot of example code as well as making my own but I have a few questions regarding both the abilities and limitations of the in-built collision detection.

I was impressed by this embedded collision detection I found here - but I see this was designed with MMF2's limitations in mind. Is the collision detection in CF 2.5 better or the same? What are the limitations of the in-built collision detection that makes this embedded solution more attractive? I did try searching the forum but I found it hard to tell what the differences between MMF2 and CF2.5 are in this regard, if any.

Also I am interested in making use of the PMO extension - is there any documentation that I can refer to that will help me out? I found this article from 2003 but does anyone know of anything newer/more extensive?

Thanks in advance!

Basically, it's for whenever the collision mask doesn't want to be the same shape as the object itself.
For example, if you have a medieval knight with an attack animation, you only want to detect when his sword touches an enemy, not the rest of his body. The alternative is to use separate invisible "detector" objects placed relative to the main sprite - but the issue then is that if you have several knights and several detectors, it can be quite tricky keeping them paired up correctly.
I doubt there's any difference at all in the way they work in MMF2 and CF.

Ahh thanks, that makes sense! I guess the next step will be for me to play around with this and see how it goes! I'm thinking about how to make animated NPC characters who just walk left/right on platforms but there's a lot of options to choose from out!