用 Sonnet 5.5 重做一版 Clawdnam Style 音乐视频
作者让 Sonnet 5.5 基于 Clawd Animation Base 重制音乐视频，14 个 agent 并行分镜、卡点到毫秒，故事由 Opus 5.5 写。
CLAWDNAM STYLE 💃🕺 I asked Sonnet 5.5 to build me a music video, and I'm having a hard time telling the difference between it and Opus 5.5. *PROMPT IN COMMENTS* I'm honestly blown away by the creativity of this model. I used @other__reality's Clawd Animation Base to create a skill that reimagines existing music videos, and puts its own spin on them. The result is kinda amazing. All the dancing, choreography, and scenes are timed to the beat. 🤯 Not only did it capture the essence of the original video, but it made some fairly hilarious changes, including a very unenthused horse, and it also included the view counter that broke YouTube's 32-bit counter for the original Gangnam Style video. How it worked: • Studied the original frame by frame and rebuilt the choreography • Measured the song's beat grid to the millisecond • Wrote a storyboard, then split the film across 14 agents working in parallel, one chapter each • Checked its own work: character sheets, every cut, every frame I don't know what Anthropic put in these models, but if this is pacing the frontier, then I'm curious what acceleration looks like. One note: Opus 5.5 wrote the story, Sonnet did the animations.