when it comes to character modeling there are a few choices the artist has to choose from when it comes to modeling preference. Not all of these techniques are best suited for the same task, they all have their pros and cons.
Box modeling is the process in which the artist starts off with a simple mesh such as a cube and slowly adds edge loops to define the topology of the character. The more loops added, the more detailed the character becomes. Artists need to be careful when adding edge loops because too many can cause a poor flow in topology.
Edge modeling is a technique were the artist outlines the defining features of the character and then proceeds to fill in the gaps. This technique allows for good edge flow because each of the primary features is built independently. Once the mesh outline is complete the rest falls into place rather effortlessly.
Digital sculpting is a type of character modeling, but not necessarily one that can be used in games due to the high amount of polygons used in the process. The artist will start off with a highly subdivided object and then used a drawing tablet as a tool to manipulate the mesh, just like using your hands to mold clay in real life.
Image based modeling is pretty self explanatory. Using image reference to directly model off saves a lot of time. The example picture given is an architectural example, but the same can be done with concept art of characters. Orthogonal images give the most information and directly correlate to each other.
3D scanning takes a tangible object and translates it into 3 dimensional space. Many people would love to use this technology but sadly isn’t available to majority of modelers. 3D scanning can also be excessively detailed and the edge flow that is generated can be messy, not ideal for animating.