The client came to me with an idea already forming, an AI-generated image of herself that captured the look she wanted, but existed only as a flat PNG. No layers, no source file, nothing usable for animation. She needed her actual likeness, faithfully preserved, turned into a character that could speak, express, and gesture naturally for her YouTube content.

The main challenge was working entirely from a flat image with no separated elements. Every part, eyes, mouth, hair, arms, had to be manually reconstructed and separated by hand before any rigging could begin. Since the reference was AI-generated, there were also hidden areas (behind hair, under overlapping features) that needed to be recreated from scratch to hold up under movement, along with careful texture matching so nothing looked pieced together once animated.

We started with a quick call to align on exact scope: facial expressions, accurate lipsync, head movement, and natural hand gestures, while keeping the lower body simple since it wasn’t needed for her format.

After delivery, the client had a few refinement requests, adjusting the chin, shortening the neck, softening some linework, and eventually asking for a second “happy” mouth set so her expression stayed warm even mid-sentence. Each of these was folded into the existing rig without needing to rebuild anything from scratch, exactly the point of building it properly the first time.

The result: a fully rigged, photo-realistic character that performs her natural speech, expressions, and gestures in real time, requiring no animation experience on her end. She now records content regularly using the system, and has already returned for additional refinements as she continues building out her on-screen presence.