Overview
In this section, we will see the algorithm overview of the TAA rendering technique.
TAA is based on the collection of samples over time by applying small offsets to the camera projection matrix and applying some filters to generate the final image, like so:
Figure 11.3 – Frustum jitter
There are various numerical sequences that can be used to offset the camera, as we will see in the implementation section. Moving the camera is called jittering, and by jittering the camera, we gather additional data that we can use to enhance the image.
The following is an overview of the TAA shader:
Figure 11.4 – TAA algorithm overview
Based on Figure 11.4, we’ve separated the algorithm into steps (blue rectangles) and texture reads (yellow ellipses:.
- We calculate the coordinates to read the velocity from, represented by the Velocity Coordinates block.
This is normally done by reading a neighborhood...
