One illustration to a full VTuber model: dev builds open-source kit using Claude and Codex agents
bdsqlsz · x · 2026-10-28
Developer lnkiai demoed an in-progress VTuber creation kit that turns a single character illustration into a working VTuber model, automating layer separation, assembly, and motion reproduction.
Notably, the workflow itself was built by running Claude and Codex as coding agents for about 5 hours of autonomous research, covering everything from layer splitting to verifying motion physics — which also yielded a surprisingly rich dataset for future training.
The kit is planned for open-source release, aiming to drastically lower the barrier to making VTuber models.
Related event: AI Coding Agents Turn Single Illustrations Into Animated VTuber Models(2 posts)→
More from coding & agent
- Solopreneur runs his company on a swarm of local-first agents, promising receipts — AIFlow_ML · 2026-10-28
- Running your AI agent from a phone daily: Telegram threads, Whispr dictation, Tailscale links — vincent_koc · 2026-10-28
- Intent-first Text2Cypher: LLM never writes Cypher, but only 1 of 4 runs correct — JeremyCMorgan · 2026-10-28
- Open-source GitHub Actions MCP Server lets agents create and manage CI workflows — modelcontextprotocol · 2026-10-28
- Dev seeks advice on wiring a custom automation agent into ChatGPT or Grok — Ok-Author-27 · 2026-10-28
- 5 mistakes to avoid when running multiple AI agents, from a practitioner — bfrench · 2026-10-28