Wan 2 7

Wraps Wan-AI's 2.7 video model through RunComfy's CLI to generate text-to-video clips with audio-driven lip sync. The standout feature is passing an `audio_url` to sync motion to your own voiceover track, which makes it genuinely useful for multi-language ad variants or spokesperson videos where you control the script. It handles 2 to 15 second clips at up to 1080p, supports negative prompts for concrete issues like flicker or watermarks, and auto-expands short prompts unless you disable it. The docs are thorough about when to route elsewhere: HappyHorse for blind-vote quality, Seedance for in-pass voice generation, Kling for editing existing footage. If you need controllable lip sync with your own audio, this is the pick.