Today, high performance image generators can be built utilizing Commercial, Off-The-Shelf (COTS) PC hardware, graphics cards and operating systems, leveraging custom software at several system levels. Image generators (IGs) based solely on COTS PC technology and custom software produce impressively powerful simulations within the COTS constraints on memory size, processor speed, processor algorithms, multi-threading, and PC graphics video outputs. This technology is being employed for fast-jet training for the F-35 “Lightning IIâ€� Joint Strike Fighter (JSF), FAA/EASA level D, ground warfare, part-task trainer, unmanned aerial vehicle (UAV) and dismounted infantry applications. Purpose-built rendering hardware also delivers impressive and powerful simulations by employing COTS Field- Programmable Gate Array (FPGA) technology to create targeted rendering solutions that exactly meet specific simulation and training requirements. Considering baseline hardware costs, these systems are expensive (today), but deliver higher quality imagery and more effective training scenarios because they are uninhibited by third party PC graphics card constraints. Today, this technology is being delivered on various devices, including those requiring FAA/EASA level D fidelity, weapons and targeting simulations in various sensor domains, and for multi-crew tactical helicopter training devices like the Apache Longbow Crew Trainer for the pilot and copilot gunner stations. PC graphics technology, largely driven by the video game industry and its variants, is here referred to as gameCOTS. FPGA technology, when delivering purpose-built image generation systems, is here referred to as simCOTS because it specifically emphasizes simulation training requirements. This paper compares and contrasts these two innovative rendering approaches to highlight the need for the simulation industry to employ a broad variety of solutions in effecting world-class training solutions, across the training spectrum, that remain squarely positioned on the cost-value curve.