Drop in two character sheets and two voice samples, then map @Image 1 to the speaker character and @Audio 1 to their voice/accent (include height), and map @Image 2 to the listener character and @Audio 2 to their voice/accent (include height).
Use a structured Script Prompt with a Global Style block plus timestamped shot beats (e.g., [00:00-00:04] Shot 1 …) to control framing, camera moves, lighting, and audio cues across the scene.
Replace all placeholders like [INSERT TOPIC HERE], [INSERT VISUAL THEME RELATED TO TOPIC], and dialogue fragments; if you leave [INSERT TOPIC HERE] blank the projected overlay topic will default to a random dark topic.
Specify lens and grading explicitly in Global Style (vintage anamorphic 35mm, cold desaturated grade with sickly greens/sterile blues, harsh low-key chiaroscuro) to lock the cinematic look.
Bake in sound design in the prompt (industrial drone loop + breathing/chuckles/gasps tied to @Image 1/@Image 2) to guide vocal performance and ambience.
Direct composition using screen thirds and shot types (over-the-shoulder medium close-up, rack-focus creep-in, extreme close-up, whip-pan, high-angle tight, macro eye shot, wide pullback) to control camera blocking and intensity.
