Learning 3D is a quick and simple series for new Sketchfab users to explore and learn about the basic principles of 3D. These tutorials will offer a foundation of knowledge and prepare you to create your own 3D models or scans on software outside of the Sketchfab platform.
Let's start from the beginning.
We know a square is a 2D object - those two dimensions being the length and the width.
A cube is a 3D object because, along with the length and the width, a dimension of depth is added.
A vertex is a point placed in the space. More than one vertex are called vertices.
With a 3D cube, for example, the vertices correspond to the corners. They define the shape.
Linking 3 or more vertices together creates a face.
- Linking 3 vertices creates a triangle.
- Linking 4 vertices creates a quad.
- Linking more than 4 vertices creates a polygon.
A face is a 2D geometry. Since a 2D geometry can be a triangle, a square, or a polygon, a 3D model is a combination of one or several 2D or 3D geometries.
But if a 3D model is made out of faces, how do we know which ones? Triangles? Quads? To find out in the Sketchfab viewer, we can look at all the edges of the faces of a model using the wireframe option in the Inspector:
Each face has an orientation defined by its vertices. An orientation data called normal is attached to each vertex. In other words, each vertex looks at one direction, and this direction is the normal.
A mathematical process called interpolation will take the orientation of each vertex and automatically calculate the orientation of each pixel of the face. With interpolation, the orientation of all the pixels between the vertices will be guessed.
Normals will be useful to manage reflection, when we add some lights to your 3D scene.
If the normals attached to the vertices are perfectly parallel between each other and go in the same direction (like the above image), the face will look flat.
But if the normals attached to the vertices look at different directions, the face could have a curved shape (like the sphere below and on the right). That’s interpolation!
Vertex normals have a big impact on how the object looks and how it reflects the lighting.
While a 3D model might look different, using normals does not actually change the shape or vertices of an object. It just gives the illusion that the shape is smooth and built from many points (vertices). It appears smoother.
Why not actually change the shape of the sphere by adding more vertices? Because performance will suffer. The speed at which you load and render a 3D model depends (among other things) on the number of vertices. Adding many more vertices to make your model smoother would impact performance quite a bit.