Cinematic in the Age of Automation: A Comparative Study of Google Veo 3.1 and Human-Directed Advertising Videos

Abstract

The emergence of AI-based video generation tools marks a significant inflection point in the history of commercial advertising production. This study conducts an experimental comparative analysis between video advertisements produced through the Google Veo 3.1 model on the Google Flow platform and those crafted through conventional pre-production, production, and post-production workflows. Taking the Coca-Cola for Everyone commercial as a benchmark, the experiment systematically reconstructed seventeen shots using AI-driven text-to-video and image-to-video generation, documenting each stage of prompt engineering, iterative refinement, and output evaluation. Four principal findings emerge: the indispensability of visual image references for replication fidelity; the limited but meaningful role of structured JSON prompting in communicating multi-dimensional cinematic parameters; the dramatic cost advantage of a fixed AI subscription relative to conventional production budgets, before accounting for the researcher’s own prompting labor; and the conditional time efficiency of AI workflows, bounded by the complexity of the target content. Interpreted through Rogers’ Diffusion of Innovation Theory and grounded in recent scholarship on generative diffusion models, visual communication, and advertising esthetics, the findings indicate that, within this specific case and this specific tool, Google Veo 3.1 reconstructing a single, product-centric commercial using visual reference images, AI-assisted video generation currently occupies a position of growing utility for studio-aesthetic, product-centric content, while retaining measurable limitations in temporally complex, choreographically demanding sequences. These findings characterize this bounded case rather than AI video generation as a general category, and the reconstruction task itself, anchored throughout by visual references, likely represents a more favorable scenario for AI performance than original, brief-driven commercial production would provide. The study contributes to the emergent scholarly conversation on AI creative authorship, cinematic quality benchmarking, and the democratization of professional video production.

The authors' abstract, as published at the source. Howard Journal of Communications, 2026 · DOI ↗