Ctuber | Scalable Character System & Rive Logic

This project focused on building a high-performance, scalable character system for Ctuber, an AI platform that generates real-time video content. The challenge was to create a digital avatar that could not only "speak" through AI-driven audio but also adapt its energy and behavior dynamically, all while remaining lightweight enough for seamless web implementation.

Working closely with the Ctuber team, I translated their character designs into a sophisticated Rive architecture. I built a custom State Machine that manages everything from idle transitions to emotional triggers.

A key part of the process was the Viseme Mapping: I designed a specialized set of mouth shapes optimized for their AI lip-sync engine. Additionally, I implemented a Modular Rigging system that allows for clothing and style changes by simply swapping skins within the same skeletal hierarchy, ensuring the character's "soul" and performance remain consistent regardless of the visual style.

The objective was to move away from static or pre-rendered video and develop a modular motion system. The character needed to support instant outfit swaps and react to AI prompts in real-time. By leveraging Rive, we aimed to reduce technical debt and provide the Dev team with a single, "smart" asset capable of handling complex interactions without the need for constant re-animation.