Reel Restyle (cross-avatar style transfer)

Take a reference avatar that has already been analyzed (e.g. lolo, with *.analysis.json + transition_style.json + subtitle_style.json), capture the style and structure of its reels as a reusable template, and re-apply that template to a brand-new avatar supplied as just a picture + voice sample + script.

The new reel reuses the reference's scene/beat sequence, talking-head vs B-roll pattern, per-beat camera angle, transitions, SFX placement, caption style and proportional pacing -- but with the new avatar's face, location, cloned voice and a different script. Actual scene durations are not copied (a different script has its own rhythm); they fall out of the new narration's word alignment, exactly like avatar-reel-composer's single-narration / hard-cut pipeline.