Black Sea Fishing Town Morning Narrative

MiniMax H3 Qwen Image 3 Pro RTX Pro 6000 96GB Local AI “Two Coffees” — 60 sec A small fishing town on the Black Sea, early morning. Two characters, one primary location — keeping the faces consistent throughout the entire sequence. Characters: Grandpa Stavri — approximately 75 years old, weathered face, short white stubble, dark blue knitted cap, thick wool sweater. Mira — approximately 22 years old, waitress, dark hair tied in a bun, green apron. Time Duration Scene 0:00–0:10 10 s Sunrise over the harbor. Fishing boats, ropes, seagulls. Stavri walks slowly along the pier with his hands in his pockets. 0:10–0:22 12 s He enters the small café. Mira looks up from the coffee machine. Stavri: “Two. As always.” She nods without asking another question. 0:22–0:32 10 s He sits by the window. He places the second cup opposite the empty chair. Close-up: rising steam, cracked hands, a worn wedding ring. 0:32–0:44 12 s Mira walks over, gathering her courage. Mira: “Grandpa Stavri… who’s the second coffee for?” A pause. Stavri: “For the one who was always late.” 0:44–0:53 9 s He smiles at the empty chair. The rising sun shines through the window, bathing the table in warm light. A seagull flies past the window. 0:53–1:00 7 s The following morning. An empty pier. Mira pours two coffees and leaves one by the window. Fade to black. Technical workflow: Create 6 keyframes with Qwen Image 3 Pro, aspect="16:9" → 1344×768. Each subsequent keyframe uses the previous one as a reference to preserve facial consistency. Generate 6 video clips with MiniMax H3 (image-to-video) at 768 px height. Clip durations are rounded to the model’s frame grid (17k+5 frames @ 24 fps), resulting in approximately 10.1 / 12.25 / 10.1 / 12.25 / 9.4 / 7.3 seconds, for a total runtime of about 61 seconds. Dialogue: ElevenLabs TTS, layered over MiniMax’s natural ambient audio with sidechain ducking, following the same approach used in “The Paper Boat.” Music: ElevenLabs Music featuring solo piano and warm strings to unify all six clips into a seamless short film. The music enters subtly at 0:20 and opens into a major key at 0:44. Editing: Use FFmpeg with 0.3-second xfade transitions between scenes.