LED 버추얼 프로덕션 가상 공간 최적화 파이프라인의 지각적 타당성과 분위기 형성 효과 분석

초록

(Background and Purpose) This study draws on Gernot Böhme's aesthetics of atmosphere and Juhani Pallasmaa's theory of peripheral vision, both of which suggest that spatial experience is constituted not by the precision of visual detail but by atmosphere and the holistic tonality of peripheral vision. Applying this premise to virtual-production background production leads to the hypothesis that lowering rendering detail and redistributing the resulting resource savings should not seriously compromise perceptual quality—a hypothesis that also addresses a problem already recognized in the Korean virtual production field: the time and resources wasted on unnecessarily high-detail, high-specification virtual backgrounds. This theory-driven hypothesis was examined through two sequential empirical experiments. (Method) Two experiments were conducted in a 28m × 6m LED-wall virtual production studio, filming the same virtual background (a daytime factory yard) under identical conditions. Experiment 1 filmed backgrounds produced with a high-spec pipeline (4K texture, standard shadow, full Nanite) and an optimized pipeline (1K texture, 75% shadow, 40% Nanite) under two lens conditions—32mm (2.5m focus distance, deep focus) and 75mm (6m focus distance, shallow focus)—to test whether changes in background detail were perceptible. Experiment 2 applied the Böhme–Pallasmaa framework to examine whether the resources saved through optimization could be effectively redirected toward atmospheric enhancement by adding a dynamic object (smoke) to the optimized-pipeline baseline footage as an atmosphere-generating element. After confirming in advance that this condition still consumed fewer resources than the high-spec baseline, the same two lens conditions were filmed to test whether atmosphere and spatial immersion improved. A fully blind online survey was conducted among video-production professionals with virtual-production experience, yielding 23 valid responses (n= 23). (Results) In Experiment 1, 86.957% of respondents under the 75mm shallow-focus condition perceived no difference between the high-spec and optimized pipelines. Under the 32mm deep-focus condition, a majority did not perceive the difference as clearly noticeable, yet more than half detected a slight difference, in contrast to the 75mm shallow-focus condition, in which 86.957% perceived no difference. In Experiment 2, the dynamic-object condition reduced project file size by approximately 45.2% and the Unreal Engine resources required to run the media server by approximately 13.4% relative to the high-spec baseline. Despite this reduction in resource use, 95.652% of respondents under both lens conditions reported improved spatial immersion and atmosphere, and 82.609% specifically cited enhanced depth and presence created by the dynamic object as the reason.(Conclusions) The perceptual discriminability of a rendering-optimization pipeline that simultaneously reduces texture, shadow, and Nanite settings is low, and the difference is particularly difficult to perceive under shallow-focus conditions. By contrast, redistributing the saved resources toward dynamic objects and other atmosphere-generating elements can effectively enhance viewers' phenomenological sense of immersion regardless of lens condition, supporting a shift from detail-centered toward atmosphere-centered workflows in virtual-production background production.

키워드