Lecture 8 Flashcards
What is rasterisation and what are some disadvantages?
Rasterisation is part of the graphics pipeline, it involves computing for each triangle what pixel is covers. Some disadvantages are: It can only be done with scan convertible primitives, requires conversion of copmlex objects into polygons, can have shading artefacts(fixed with per-pixel shading), does not have automatic handling of shadows, reflection, transparency.
What does the graphics pipeline do?
It takes in a geometric model made up of vertices, normals, and texture coordinates, and a lighting/shading model(light positions and view point), and outputs a Colour and depth for each pixel on screen.
What are the parts of the graphics pipeline? Which parts are done in the CPU and which in the GPU?
The application sends commands to the vertex processor, which sends vertices to the primitive assembler, which sends primitives to the rasterisation, which sens fragments to the fragment processor, which sends the pixels to the display.
Only the application occurs on the CPU, the rest is done on the GPU.
What are the parts of the vertex processing?
Vertex shader provides coordinates in clip space to clipping, which sends clipped coordinates to projection, which provides the coordinates in the image plane, which is sent to the viewport transofrm which outputs the window space.
What is the vertex shader? What occurs in the next phase?
Transforms the input vertex stream into a stream of vertices mapped onto the screen(clip space coordinates), this is done using the model, view and projection matrices to transform from model to world to view and to clip space.
In the next phase clipping is done, removing coordinates that are not in view.
What occurs in the projection and viewport transform stage of vertex processing
projection transforms the clip space coordinates into normalised device coordinates.
Viewport transform mpas the resolution independent normalised device coordinates to a rectangular window in the frame buffer.
What occurs in primitive assembly?
Receives processed vertices and vertex connectivity information, it divides them into a sequence of base primitives(primitive generation) and processing), these will face a certain way, the back face is then culled in face vulling.
What occurs in fragment processing?
Takes fragments from rasterisation and processes them to add color and depth values.
What is OPENGL? How does it work?
A platform independent programming interface for interacting with graphics processing. It works by using a client-server model, the application/client creates commands and the server processes commands(can be same machine).
What are the three steps in rendering an AR scene?
- render video texture.
- Calculate camera pose.
- Render virtual data with camera pose.
What is an OPENGL context? How do these interact?
An instance of OPENGL, stores all state of that instance. Processes can have multiple of these representing a seperate viewable surface., these can share resources. Contexts can be current for a thread with one to one mapping, and operations work on the current context.
