Epiphanic resolution: The effect of video compression on the believability of computer-generated characters
Researchers: Jason Kennedy
This article examines the effects of video compression on the believable integration of computer-generated (CG) characters among live-action film elements. Compression is requisite for the delivery of moving-image content to a variety of end-user applications, including cinema, online streaming, Blu-ray and digital video files. The most common standards for compressing consumer-targeted video content include H.264/Advanced Video Coding (AVC) and H.265/High-Efficiency Video Coding (HEVC), each of which provides separate pros and cons depending on the type of footage and the degree of compression required. This research investigates a previously unexplored question: to what extent does the type and degree of compression affect how well virtual actors (vactors) appear to coexist within profilmic scenes? By extension, what visual results linked to compression have the greatest impact on compromising a vector's believable integration within a shot? Analyses of two feature films and a web-based advertisement at various compression strengths provide data that strongly suggests that compression is more detrimental to the believability of CG vs. profilmic characters within the same shot. Additionally, as compression strength increases, CG characters become more graphically abstracted, negatively impacting the quality of their visual integration, whereas profilmic actors remain recognisably human and plausibly integrated – a phenomenon I dub ‘epiphanic resolution’. This research provides novel insights regarding the relationship between the finished video product as delivered by a film company vs how it may be perceived when viewed in different formats by a variety of audiences.
