Camera and OpenGL

In OpenGL: no real notion of "camera"

- Unique viewpoint - cube \((x,y,z)\in[-1,1]^3\) - Normalized Device Coordinate
- Left-right/x, Bottom-up/y, Front-back/z (toward negative z)
- No perspective (displays (x,y) for the smallest z)

Any camera effect (position, orientation, perspective)
- must be coded "manually" in the shader.

- All these effects can be represented as matrices.

Three common matrices
- Model matrix

- - Affine transform - specific to a shape.
  - Local object coordinates to world space.

- View matrix

- - Rigid transform - position/orientation of the camera in the scene.
  - World space to camera/view/eye space

- Projection matrix

- - Perspective matrix - map frustum to normalized cube. (/or ortho projection)
  - Camera space to Normalized Device Coordinate (clip/screen space)

- \(p = Projection \times \underbrace{View \times Model}_{ModelView} \times p_0\)

assets/view_space.png