This document discusses optimizing a photo editing application called PhotoDirector to take advantage of AMD's heterogeneous system architecture (HSA). It describes how photo editing pipelines involve computationally intensive RAW processing that could benefit from GPU acceleration. HSA allows sharing memory between the CPU and GPU to reduce bottlenecks. Performance tests show the potential for a 2x speedup using coarse-grained shared virtual memory buffers over OpenCL. The document concludes that HSA has great potential to improve performance for parallelizable and memory-intensive tasks in photo editing applications.