Descriptions:
Craig Hewitt walks through a complete workflow for creating an AI video clone of himself using Higgs Field — a platform he describes as the “OpenRouter for generative media” — which aggregates image, video, and audio models including Seed Dance, Kling, ElevenLabs, MiniMax, and others. The tutorial centers on Seed Dance 2.0 4K, which Hewitt identifies as the best currently available model for talking-head video with synchronized voice.
The process involves scripting via Super Whisper transcription, uploading a voice sample for cloning, providing a reference photograph, and writing detailed scene prompts that specify naturalistic behaviors — phone camera bobbing, gaze shifts, walking a dog — to prevent the mechanical stillness that marks AI-generated video. Hewitt discusses Higgs Field’s token economy, recommends generating voice separately in ElevenLabs for more natural intonation, then feeding that audio back into Higgs Field for lip-sync rendering. He also notes the platform’s MCP and CLI support for agentic workflows via tools like Claude Code or Codex.
Beyond the technical walkthrough, Hewitt frames the video around a genuinely open question for content creators and marketers: as AI-generated video becomes indistinguishable from human-produced content, is leaning into it a strategic advantage or a brand risk? He intentionally leaves that question unanswered, making the video useful both as a practical guide and as a framework for thinking through the decision.
📺 Source: Craig Hewitt · Published July 23, 2026
🏷️ Format: Tutorial Demo
