Virmii: Turn one character image into a live 3D VTuber

Hey Product Hunt! 👋 I built Virmii after working on a native macOS avatar overlay and repeatedly running into platform-specific permissions, packaging requirements, graphics APIs, and a workflow that was difficult to bring to Windows and Linux. Virmii moves the entire path from character idea to livestream into the browser. Start with one reference image, generate and prepare a rigged VRM, preview it, use camera-based face tracking, and open a private transparent stage in OBS. You can also download the VRM for use in compatible tools. One principle shaped the product: generated models do not all have the same facial topology. Virmii validates each result and clearly distinguishes full facial tracking from a reliable head-and-body fallback instead of pretending every output is identical. I’d love feedback from VTubers, streamers, and avatar creators—especially about the reference-to-live workflow and what you’d like us to improve next. Thanks for checking out Virmii! 💜