[Overall Style] Cinematic deadpan music comedy, commercial film quality, designed for Seedance 2.0's multi-reference character locking, multi-shot continuity, and native audio-visual sync for Mandarin singing and dialogue. The illusion of a grand immortal concert collapses into an awkward private KTV room.

[Character A] 25-30yo East Asian 'Sword Fairy' from @Image 1, strictly maintaining face, eyes, long black hair, white embroidered silk Hanfu. She stands on a glass table holding a silver sword like a microphone.
[Character B] 'Bike Sister' from @Image 2, maintaining face, ponytail, and position on the sofa.

[Shot 1 | 0-5s | 24mm Wide Low Angle] A luxury private KTV room with mirrored walls, red velvet sofa, fruit platter. Character A stands on the table holding a sword while Character B watches. Sword qi particles gather as epic strings swell.
[Shot 2 | 5-10s | Medium Shot] Camera circles. Character A closes eyes and belts out an exaggerated Mandarin opera high note. Hair and sleeves float, glasses vibrate, and the KTV screen turns into a cloud sea. Character B looks genuinely shocked.
[Shot 3 | 10-15s | Close-up] The high note ends, silence for half a second. The KTV system announces: "Score: 12 points." The cloud sea collapses into cheap disco lights. Character B claps twice and says: "Sister, plenty of sword qi, not enough breath." Character A lowers her sword, her face twitching as she tries to maintain a cold smile. Freeze on her broken expression.

[Technical] 15s total, 16:9, 24fps, accurate lip-sync, no identity drift.