Near-infrared (NIR) cameras are a key element of a new generation in-cabin monitoring system and they are called to become an essential piece for DMS and OMS development. However, collecting data (including NIR) for the different in-cabin use cases is challenging and in some cases impossible due to privacy issues and concerns.We all agree that synthetic data can complement real data, or be a viable alternative to real data to train perception systems. However, classical approaches to simulate NIR data rely on visual tricks. Is a faithful physically correct NIR simulation possible? We think it is and we will show how hyperspectral rendering enables the simulation and generation of accurate NIR data for in-cabin use cases.