AI Influencer: Chloe is an ongoing AI-assisted character project centered around the creation of a realistic virtual influencer. After exploring AI tools through commercials, cinematic sequences, UGC-style content, and dialogue-driven animated shorts, I wanted to build a project that could function as a long-term research platform for studying the strengths, limitations, and practical applications of current generative AI workflows.

​

The project began with a simple question: how convincingly can AI tools create and sustain a realistic human persona across images, videos, social media content, and different production scenarios? To explore this, I designed Chloe as a virtual influencer and created dedicated social media accounts for her on Instagram, TikTok, and Douyin. By presenting the work in real social media environments, I hope to observe how audiences respond to AI-generated imagery and video, and to gather feedback on what feels believable, what feels artificial, and where the current limitations still appear.

​

The primary goal of this project is to explore how AI tools can be used to construct a human-like digital identity that is difficult for viewers to immediately recognize as AI-generated. This involves not only generating attractive individual images, but also maintaining consistency in facial features, body proportions, expressions, styling, lighting, and camera language across a wide range of content. In this sense, Chloe is both a creative character and a technical test case for AI-driven identity design.

​

A major part of the project focuses on comparing different image-generation tools, including Nano Banana Pro and GPT Image 2. I use these tools to create character sheets, expression sheets, style frames, fashion references, and social-media-style portraits, while studying how each system handles facial consistency, realism, posing, lighting, and visual detail. Through this process, I am able to better understand where each tool performs well, where it struggles, and how reference preparation affects the final result.

​

For video generation, I am also testing tools such as Seedance, Kling, Sora, and Wan across different formats, aspect ratios, character actions, and visual themes. Because influencer-style content often depends on subtle body language, facial expression, camera presence, and natural movement, this project gives me a practical way to evaluate how each model performs under specific creative conditions. Some tools may be stronger for cinematic movement, while others may be more effective for social-media-style shots, close-ups, fashion content, or short-form video pacing.

​

In the editing and finishing stage, I use Premiere Pro, CapCut, Topaz Video, and ACE Studio depending on the needs of each piece. Premiere Pro is used for more controlled editing and sequencing, CapCut is useful for social-media-oriented finishing and subtitles, Topaz Video supports upscaling and image enhancement, and ACE Studio is used when music or sound elements are needed. This allows me to treat the project not just as image generation, but as a complete AI-assisted content production pipeline.

​

Since AI Influencer: Chloe is designed as a long-term project, it will continue to evolve as new tools, models, and workflows become available. For me, this project is both a creative experiment and a practical research process: a way to study how AI can be used to build believable digital personalities, where the current technology still falls short, and how artists can guide these tools with stronger visual direction, production planning, and post-production decisions. It also serves as an open invitation for audience feedback, helping me better understand how people perceive AI-generated human characters in real online environments.