Is creasing considered a good modelling practise?

Just curious. Most 3D programs have the ability to weight or crease vertices or edges, essentially creating sharper edges without adding extra topology (on subsurfed models). For the hardcore modellers out there, is this actually considered good or reasonable modelling workflow? When I was taught modelling I was taught to add extra topology to create sharper or smoother geometry.

Cheers

I think it strongly depend on what the model will be used to.
For rendering it on a still picture or an animation, if the end result is satisfying you can keep using crease if you want.

Now for exporting it into some engine (game or other) that use models in a real time environment, it will depend a lot if the format used in such engine support crease or edge marked as sharp , or not .

If the format of model used in such engine do not support crease or edges marked as sharp, you will need to add additional topology yourself. If the format use crease or edge marked as sharp, i think it’s then best to -not- add additional topology as you would then increase the face count a lot more than required.
A real time engines, like games, are all about face count optimisation (as it must be able to display environement+ model etc… in the same time, in the end adding all the faces of everything that can displayed in the field of view of the user can be high)