GitHub radar
Turn one photo and a script into a talking digital presenter video
This is a ready-made skill for your AI agent: feed it one photo and a script, and it builds a finished video with a talking digital presenter, start to finish.
A task template (skill) for Claude Code or another AI agent: give it a photo and a script, it creates a complete talking-head video with a digital presenter.
When it helps
Useful for a quick ad or explainer video with a presenter, without filming anyone; skip it if you want something free or fully local — generation runs in the cloud and costs credits.
Pros
- Agent writes the script, edits the video, and adds captions itself
- Shows the exact credit cost and asks for confirmation before running
- 5 ready editing styles plus 48 packaging elements included
- No GPU needed — all generation happens in the cloud
Cons
- Requires a paid Renoise account and credits for every video
- Still needs local ffmpeg, Python, and Node.js installed
- The text-behind-presenter effect only works on macOS
How to set it up — step by step
- 1Open your AI agent (such as Claude Code) inside your project folder.
- 2Send the agent: install the digital-human-koubo skill from github andyhuo520 per its README and connect Renoise MCP.
- 3Create an account at renoise.ai and log in through the browser when asked — no API key needed.
- 4Check local tools by running bash ~/.claude/skills/digital-human-koubo/scripts/doctor.sh.
- 5Send the agent the text from the block below.
- 6Review the finished video and confirm the credit cost before the final run.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install this skill from https://github.com/andyhuo520/digital-human-koubo into my agent, and make a talking-head video from this photo [photo/description] and this script; ask me for the Renoise account and credits if you need them.
Video and voice generation runs in the cloud through the Renoise service, not on your machine — you need a separate account and paid credits (the README's 27-second clip cost 482 credits). Locally you only need ordinary ffmpeg, Python and Node; no GPU required. The text-behind-presenter effect works on macOS only, since it uses Apple's Vision framework.
Open on GitHub▌ More finds