Race-Making in Real-Time 3D

Reality is being captured to automate the work of creating 3D game worlds. Framed as more efficient than traditional 3D modelling, these ‘Reality Capture’ techniques extend to human 3D scanning that use photogrammetry to match up points of interest across multiple scans of physical objects, compositing scans into 3D meshes for computational manipulation. While the dominant aspiration for human beauty on social media is smoothness and symmetry, the benchmark for quality in digital humans is texture and irregularity. Within the visual economy of hyperreal videogames, imperfection imparts ‘character’ to artificial beings, enhancing their capacity to engage, influence, and entertain. Requiring more resources to model, visual character can now be synthesised from databases of diverse facial features, skin tones, and hair textures captured from photogrammetric scans of real humans. Predominantly white makers of digital human tools such as Epic Games’ MetaHuman Creator promote how blending from their ‘diverse series of presets’ empowers equitable character creation. Scholars of ‘post-race’ call out these practices as diversity surfing, which casts race as a cosmetic rather than political category, producing compliant digital humans who stand in for black and creators of colour. Building on critical race studies of biometrics and facial recognition technologies, I analyse how digital human tools misrecognise the mathematical facticity and optical veracity of photogrammetric databases. These tools disaggregate the lived experience of race and the fraught politics of diversity into statistical variability. This talk argues that this form of race-making – and unmaking – is attributable to game engines as a software framework. Just as game engines partition base systems such as physics and rendering from gameplay modules, photogrammetric tools extend an understanding of race that segregates core technologies from peripheral assets: the control rig of human kinesiology is insulated as proprietary function, while the ‘ever-growing library of variants of human appearance’ is relegated as fungible asset. As game development incorporates generative AI techniques that increasingly rely on photogrammetric and synthetic data, the stakes of game engines go beyond representing race in games to shaping cultural understandings of racial coherence, order, and difference.