In audio-visual augmented and virtual reality applications, the audio delivered must be consistent with the physical or virtual environment, respectively, in which the viewer/listener is located. Artificial binaural reverberation processing can be used to match the listener’s/viewer’s environment acoustics. Typical real-time artificial binaural reverberators render the binaural room impulse response in three distinct section for computational efficiency. Rendering the response using different techniques means that within the response the early reflections and late reverberation may not give the same room-acoustic impression. This paper lays the theoretical foundation for early reflection remapping. This is accomplished by acoustically characterizing the virtual room implied by the early reflections renderer and then later removing that room-character from the response through frequency-domain reshaping.