Graphics Pipeline

Graphics Pipeline Stages

1. Vertex Array

2. Vertex Shader

3. Triangle Assembly

4. Rasterization

5. Fragment Processing

6. Frame Buffer

Vertex Shader

Operates on each vertex to interpolate values and apply shading (e.g., Gouraud shading).

Triangle Assembly

Connects vertices to form triangles for rendering.

Rasterization

Converts vector format (triangles) into a bitmap or grid of pixels. Clips parts outside the screen.

Fragment Processing

Calculates color and depth for each pixel fragment, receiving input from the vertex shader.

Frame Buffer

The final destination for rendering output.

GPU

Reads selected vertices from the vertex array, runs vertex shader to form triangles.

Rasterizer

Clips and breaks triangles into fragments.

Frame Buffer (Vocab)

The final output destination for a rendering job.

Digital Differential Analyzer (DDA)

Algorithm that helps determine the best pixel filling for a line between two points.

Gouraud Interpolation

Color interpolation between vertices with smooth shading ; vertices have their own colors C0, C1, C2 

Culling

Removing invisible geometry to improve processing time. Includes volume, occlusion, and backface culling.

Volume Culling

Removes objects outside the camera's view.

Occlusion Culling

Removes objects blocked by others closer to the camera.

Backface Culling

Removes the backsides of objects that face away from the camera.

Clipping

Precisely removes parts of primitives outside the view volume.

Cohen-Sutherland Algorithm

Divides 2D space into 9 regions using the view volume boundary and outcodes.

Sutherland-Hodgman Algorithm

Clips polygons against each edge of the view window.

Aliasing

The jagged appearance of lines and edges. Anti-aliasing smooths these jaggies by increasing sample rate.

Box Filtering (Anti-aliasing)

Averages pixel values to reduce jaggies.

Super Sampling (Anti-aliasing)

High-resolution rendering followed by downsampling to smooth out aliasing.

View Space

Coordinate system based on the camera’s position and orientation.

Local Coordinate Space

Each object has its own coordinate system, allowing for movement relative to each other.

World Coordinate Space

All objects transformed into a single coordinate system for consistent positions and distances.

Canonical View Volume

Cube with sides of length 2, centered at the origin. Only objects within this cube are visible.

Window Space

Coordinate system matching screen coordinates with the origin at the lower-left corner.

Transformation Matrices

Used to move objects between coordinate systems.

World Matrix

Moves individual models from local space to world space.

View Matrix

Moves objects from world space to the camera's perspective in view space.

Projection Matrix

Projects 3D objects onto a 2D screen. Includes orthographic and perspective projections.

Orthographic Projection

Objects retain their true size and shape without distortion, used in CAD.

Perspective Projection

Mimics how objects appear smaller as they move farther away, creating a realistic view.

Specifying a View

Origin (0,0,0) and is the position of the camera, gaze direction is where the camera is looking, positive y axis points upwards, positive x axis points to the right