Interview with an AI Actor is an AI-generated animated short film built around the format of a recorded interview with a virtual performer. In my previous AI-assisted projects, I had explored 3D commercials, photorealistic advertising, cinematic sequences, and UGC-style content. However, this was my first attempt at creating a dialogue-driven 3D animation using AI tools. I chose this format because an interview naturally emphasizes performance—especially facial expression, timing, and lip-sync—which makes it a strong test of how well current generative AI can handle character-driven storytelling as of 2026.

​

To begin, I developed the script and shot structure through iterative discussions, refining the tone, pacing, and conversational flow of the interview. I then used Nano Banana to generate character sheets and style frames, producing a large number of reference images to guide the character’s appearance and, more importantly, to stabilize facial features for dialogue-heavy scenes. These references were essential for maintaining visual consistency and supporting believable mouth movement during speech. With the visual foundation established, I used Kling 3.0 Omni to generate the video footage for each shot, and ElevenLabs to create the character’s voice performance.

​

Unlike my previous projects, this piece required a much stronger focus on lip-sync and performance clarity. Because the film is built around an interview format, the character speaks continuously, making mouth animation and timing far more critical than in action- or montage-driven work. To explore this further, I experimented with two distinct animation styles: a more conventional 3D performance approach and a stylized, exaggerated animation inspired by Pixar-like acting. Through the use of off-screen “director” voice cues within the narrative, I was able to blend these two styles in a self-aware and playful way, turning the stylistic shift into part of the storytelling itself.

​

Once the footage was generated, I edited the sequence in Premiere Pro, integrated the voice audio, and finalized the pacing. I then used Topaz Video to upscale the final output and CapCut to generate subtitles, ensuring clarity and accessibility.

This project served as both a creative exploration and a technical experiment. It demonstrated that AI tools can already support dialogue-heavy animated content when supported by strong reference preparation and careful shot planning. At the same time, it revealed ongoing limitations—particularly in maintaining character consistency over longer sequences. I found that breaking the film into shorter shots significantly improved visual stability, making shot segmentation a key strategy when working with current AI video-generation systems. For me, Interview with an AI Actor represents a focused investigation into AI-driven character performance, and a step toward understanding how these tools can evolve to support more sophisticated animation workflows.