I will compare how two character meshes deform in animation. To make this comparison mean something, I have selected one of the best base meshes I could find without directed edgeflow. This mesh is made by unknown person. It has nice even division of polygons – good for sculpting. Second mesh, seen on the right, is mine and built for both sculpting and animation. It is almost exactly the same size and shape as the first. I have rigged both meshes in Messiah with one rig – they both do exactly same motions. I haven’t done any weighting of bones to the mesh – Messiah bones have a good effect on the mesh by default. Point is that with this setup the only difference you can see comes from the meshes.
Here are the meshes in rigging pose. My Edgelooped-mesh has different head and no toes as I was lazy and many characters will be wearing shoes anyway. The edgelooped mesh has about the same number of polygons in the body as the ‘normal’ mesh, but more definition because the flows define shape. The flow also helps maintain shape in extreme motions, like seen in the stretching example. Observe the general form, especially upper shoulder and chest area and the hip. See how the edgeflow helps to keep the shape and how it deforms it a bit better? Difference is not notable everywhere, but it is there and it is important.
You may argue the first mesh would show the same definition if we just pushed points around and added a few polygons. But that’s just it – unless you add those polygons as carefully placed loops, you will have to add a lot more than ‘a few’ to get the same definition edgelooped mesh has with less polygons.
Some might also say that the flows don’t matter that much in animation production, because when final mesh is subdivided to gazillion polygons at rendertime there will be more than enough for joints and to keep the definition. I disagree. Base mesh is the one that gets animated, it sets base grid for the final – any problems in the base are still present in the final. And I dare say they become more visible in a highly detailed mesh.
Last examples show how the edgeloops help in joint areas. With the knee I’m using the loop shown here (see image with lots of loops). Same works at elbow and at shoulder-top. The loop ‘binds’ the parts together and provides material for both sides of the outward bending limb – keeps the volume.
With fingers I’m using a simpler ‘loop’ to keep polycount low. It adds one more edge on the out bending part and helps to keep the volume. It also introduces triangle-polys. If triangles are a problem, add a full around the finger loop instead.
Mesh with evenly spaced polygons does well in animation and a mesh with planned edgeflow does even better. No suprise there, but I needed to test it anyway.
What’s your take on this? Is edgeflow really that important? What is your preferred flow – care to show it? I know mine is just one way to do it.