Basics

The process of superimposing a grid over a vector image and determining the colour depth for each pixel.

In short, turning 3D to pixels.

Different ways to connect vertices in rasterization, usually to form multiple triangles

A sampled position on a surface, containing the data required to shade it.

A pixel may contain multiple fragments.

Blocks of memory that hold sprites pixel by pixel, as bits of graphical data.
Good for 2D, doesn't scale well though.

A method made up of mathematical functions that define lines, points, and curves between two points, which are taken into account when rendering the final image.
Scales better.

An advanced rendering method that is used to create incredibly realistic lighting effects for things like video games and movies.
Simulates light rays, shot from either a light source or from the camera.
Used for photorealism, very resource intensive and slow.

Vertex Operations: (Input of Vertex Attributes, Position, Colour, Texture Coordinate, Normal, etc.)
-Vertex Specification
-Vertex Transformation
-Vertex Clipping
-Perspective Divide
-Viewport Transform
-Primitive Assembly
\/
Fragment Operations
-Rasterization
-Sample Processing
--Depth Test
--Alpha Blending
-Write To Framebuffer
(Outputs Fragment Colour & Depth)

Determines the vertex setup for the rest of the pipeline by defining the memory contents (the data being displayed) and layout

Takes the vertices, currently local to the shape, and transforms them to be local to the camera view
Takes advantage of vertex reuse (a single set of vertices are made and copied for all assets)

Any shapes outside of the camera are either culled (removed entirely) from the pipeline, or clipped (partially removed)

Vertices are transformed again to add the effects of the perspective to the vertices

Vertices are transformed again so their positions equate to exact positions on screen

Individual vertices are grouped according to the primitive type, allowing for appropriate rasterization

Rasterization produces samples (fragments). Each sample includes the data that needs to be written to different buffers.
This step prepares that data for writing to buffers.

The way the Z coordinate points
Left Handedness: pointing away
Right Handedness: pointing towards

Buffers that store the results of fragment processing before they're displayed on the screen.
Tend to have two frame buffers to stop unfinished frames being displayed (Double Buffering)

Used to interact with the GPU in a simple way. These add extra wrapping to Vertex Transformation and Sample Processing, using shaders as well for these bits.
E.g., DirectX, OpenGL, Vulkan

A queue of API calls to the GPU (FIFO) which the GPU tries to get through ASAP. Allows for the GPU to work asynchronously.