it uses math to guesstimate what an image would look at a bigger resolution based on the motion vectors (the direction where object are moving to).
Its limited in scope and results compared to other solutions simply because of the comparative simplicity and lesser resources needed to run the algorithms; Other solutions are leveraging separate processors capable of using matrix operations for their calculations, while AMD/FSR chooses to do all the work on the shaders that also work on everything else at once.
Latest Answers