We'll discuss porting After Effects to run on the GPU. We'll start by presenting our reasons for starting the work along with our initial expectations. Next, we'll discuss the inherent challenges involved in attempting such a task as it pertains to the GPU in particular with an older application and architecture. We'll discuss in greater detail how we achieved performance gains in certain areas along with cases where achieving better performance was more difficult and why. We'll also incorporate examples of how we used NVIDIA-specific tools like NSight to help optimize and diagnose issues with our image processing kernels. Those with an interest in both higher and lower level architecture for a mixed CPU/GPU based image processing pipeline should attend.