This time, I don’t get into trouble, just want to hear your opinion and some explanation
As a habit, I’ve never applied modifiers, just leave them there in the stack and hit render. Applying modifiers make no difference, doesn’t it ? I only apply those modifiers before exporting the model to another program.
I just discover that Edge Split modifier adds more vertices to the model, so if you choose to apply Edge Split, you will find out your model is full of duplicate vertices, and if you choose to remove the them (remove doubles), the edge split effect is gone. I don’t really understand this, what relationship between adding duplicate vertices and Edge Split ??? I hate duplis.
It really depends on what modifiers you are using and what you are trying to accomplish. If for example you were using the array modifier to generate some bolts across a hatch the bolts would all be identical:
In a final image this is generally not ideal as in real life when a series of bolts is tightened it is extremely rare to have them all rotated the exact same. So what you would do is apply the modifier and give each bolt a bit of rotation so as to give them some variation.
The edge split modifier is generally used on low poly models. It rips edges apart from one another, the same as if you had used the rip tool (v) and kept the vertices in the same place. This done because when to face are at a high angle to each other blender has a hard time averaging the shading between the two faces (http://www.guerrillacg.org/home/3d-polygon-modeling/smooth-shading). So the edge split modifier separates the face to help correct this.
You can try this yourself by setting the shading on the default cube to smooth and then applying the edge split modifier to it.
Applying modifiers makes a difference in many cases especially if you plan to UV map and paint your models.
There are obvious modifiers that MUST be applied to complete modelling e.g. the Boolean modifier, mirror modifier.
Also if you don’t apply subsurf before painting textures on a model it distorts the seams.
Edge split really is the same result as if you selected part of a mesh and separated it with “P” - a new part that just happens to match the other part exactly. If you are working only in Blender there are better ways to get a good surface.