Best practice for facial topology

Facial Topology is the study of the face, focusing on volume, shape and proportions of the features while understanding where and how they are positioned on the face. Proportions are height, width, depth, size of one area with the size of another and the space between two areas or features.
Here are some common words used within facial topology and their meanings.
The mesh is a wire like framework that surrounds a 3D model. It is a grid of lines made up of different shapes (usually squares) that is used to determine the location of polygons when increasing or decreasing the polygon count. It also shows how the face would be animated by the direction or flow of the lines on the mesh. When the polygon count increases, a square would divide into four equal sized squares. The higher the number of lines on a mesh, the smoother the model can be, however it also takes up more memory and could slow down the computer, this is directly affected by the number of lines, the size of the model and the computing power.
A quad is where four lines on a mesh meet. It is now the most common shape on mesh as it is the easiest shape to divide when increasing the resolution or subdividing the mesh.
A tri is where three points on a mesh meet. It is usually used when the mesh flow changes direction.
Poles are where three or more mesh lines meet at a point. Poles act like anchor points which controls the flow of the mesh (direction of the mesh lines). Poles are best placed on the face where there will be the least amount of animation. There are usually around ten poles in a face model positioned in the edge loops. There is no right or wrong place to put poles, it depends on what parts of the face will be animated and what shape the face will be. Every head, whether it be characteristic, realistic, alien or monster, all heads look different, for this reason there is no universal right or wrong place to put poles.
Loops or Edge loops are loops of mesh that go around the features of the face you want to animate. They are usually highlighted in mesh pictures with different colours, this is purely to highlight their location and it has no other relevance to animation. It is important to have edge loops around features as it enables a more natural animation movement. Some edge loops mimic the facial muscles around features like the mouth and eyes; they both have circular muscle to control the movement. The edge loops are a simplification of the complex muscle network within the face, the more accurate position of edge loops to real muscle structure, the more realistic the animation.
Important things to remember and different ways to model in 3D.
There are different methods to use when modelling a 3D head; some professionals prefer a new mesh design for every face, with specific edge loops and poles created specifically for one face, others prefer to use a generic head mesh with edge loops and pools already included, which can be manipulated to fit the face. There are benefits and drawbacks to both, and neither is right or wrong, it depends on how the modeller prefers to work and what they find the most suitable for a given task.
When creating a new mesh to work from, one way to achieve this is to work from two photos or drawings of a face, one in profile, and one straight on. It is then possible to draw on the straight on photo/drawing, the edge loops in 2D. The 3D mesh is created by dragging the mesh lines out along the profile picture/drawing to get the depth of the 3D mesh.
An important thing to remember and keep in mind while doing a 3D head model is that the head is organic. The human head is a complicated combination of muscle, skin of different colours and textures, nerves, bone and hair of different thicknesses and direction. Modelling is in essence, a simplification of these complicated systems (you only have to model what’s on the outside), but understanding how the muscles, tendons and ligaments work on the inside makes modelling easier to do. It’s like you have been given a task to draw a cityscape with lots of buildings, but you own most of the buildings in the city so you have prior knowledge of where the buildings are in relation to each other.
Certain parts of the face act as reference point for others on the face, and they are somewhat universal for all heads, however, every head is different so the relation between features is only a rough guide.
The eyes are positioned roughly half way down the face; the eye length should be the same distance between the eyes, and between the eye and the side of the head. (The eye lid is usually thinker than is first thought, a thick eyelid looks more realistic than a think eyelid.)
The top of the ears, should line up with the eyes, and the bottom of the ear (lobe) should line up somewhere between the bottom of the nose and the top of the mouth.
The corners of the mouth act as reference point for several parts of the face. They line up with the centre of the eyes and create a reference point for the cheeks and nostrils.
The human brain is very adapted in recognising patterns and symmetry, the same applies for the face, if the face is made identical on both sides, it is inaccurate as no head is symmetrical. For a very accurate 3D head, the modeller has to change the shape and size of some features, eyes are not always the same size or the nose may be bent to one side.
Conclusion
There is no best practice for facial topology across any software platform, there is only best practices for the specific job you are doing, what may be a good technique in one face may be a different in another type of face. With technology advancing and the software used to create models via a computer are becoming more advanced, the techniques used to do so will no doubt change and adapt with it. The one thing that will stay the same is that all heads look different but they are share similar proportions.