The goal of this test was to continue exploring continuity in AI-generated video, specifically how to create more coherent sequences across shots and actions. A major focus was testing first-frame and last-frame workflows to see how much control they could provide in shaping transitions, preserving character consistency, and guiding motion from one beat to the next. The project also served as a way to push the limits of 3D-style animation inside an AI pipeline, evaluating where stylized motion, product interaction, and camera choreography could hold up — and where they still began to break. Overall, the intent was not just to make a finished ad, but to stress-test the system and uncover practical methods for building more controlled, cinematic sequences.

The entire background track was generated using Suno, making this ad fully AI-produced end to end.

The testing showed that continuity improves when motion is simplified and the camera does more of the work. First-frame and last-frame setups were useful for establishing direction, but they still required careful shot design to avoid broken transitions, warped product handling, and inconsistent character detail. Product-focused moments, especially hand interaction and can-to-mouth contact, proved to be more fragile than wider walking or pose-based shots. Surveillance angles, shelf POVs, and controlled push-ins held together better because they reduced complexity while still feeling cinematic. The biggest takeaway was that AI performs best when the sequence is designed around controlled beats, minimal subject motion, and strong visual anchors rather than trying to solve too many moving parts at once.

The tests also made it clear that AI-driven 3D-style animation is still not fully there yet. While some shots suggest the look and motion language of 3D, the output is not consistently polished enough to fully replace a true 3D pipeline when precision, refinement, and clean product interaction matter. It can hint at the aesthetic, but it still breaks under closer scrutiny.