Episode 2: One clip is easy. A full story is where AI video gets real → Keep characters consistent across shots → Direct scenes, not prompts → Build a narrative that holds together

Episode 1: Every great AI video starts before the first generation → Build your references → Set the visual language → Generate shots that already belong to the same film

Magnific Spotlight in San Francisco with MiniMax H3 → Showcase of projects the community made with H3 → Fireside chat: MiniMax × Magnific → Live workflow demo → Food, drinks, and networking at our SF office Spots are limited. Request yours 👇

Episode 3: at the pub Prompt: "Comedic multi-shot sketch. No music, natural sound and dialogue only, all in English. All nine stylized 3D clay-render characters from the reference sheets appear together in every shot. SHOT 1 [00:00-00:07]: All nine slouched along a western saloon bar in standoff silence, boots on the footrail, one hand tapping the wood. Black cap sunglasses man, deadpan: "Are we ordering?" Slow dolly along the bar. SHOT 2 [00:07-00:14]: Cut to the bartender's hands sliding nine tall bright pink milkshakes onto the counter — whipped cream, curly straws, tiny paper umbrellas. Black cap sunglasses man lifts his shades: "That's not what I ordered." Off-frame bartender: "Yes it is." Wide framing, all nine lean forward. SHOT 3 [00:14-00:20]: Cut to all nine drinking in grim synchrony, one long absurd slurp, then setting the glasses down together, pink foam on their moustaches. [pause] Beige cowboy hat man, quietly: "It's good." Leather cap man, without turning his head: "Shut up." Static wide shot, everyone stares straight ahead. Stylized 3D animation, warm poster-like color grading, cinematic lighting, expressive comedic timing."