
2x2 grid, 16:9, do this for 4 famous scifi movies: class UniversalEditorialMetaphor: def __init__(self, topic): self.topic = topic self.semantics = infer_semantic_structure(topic) self.thesis = infer_visual_thesis(topic) self.mood = infer_editorial_tone(topic) self.metaphors = generate_candidate_metaphors( topic, count=12 ) self. best_metaphor = self. select_best_metaphor() def select_best_metaphor(self): return max( self.metaphors, key=lambda m: 0.35 * semantic_accuracy(m) + 0.25 * conceptual_originality(m) + 0.25 * visual_clarity(m) + 0.15 * silhouette_strength(m) ) def visual_actors(self): return { "primary": infer_dominant_form( self. best_metaphor ), "secondary": infer_consequence_or_partner( self. best_metaphor ), "interaction": infer_decisive_action( self. best_metaphor ) } def composition(self): return { "geometry": "asymmetric", "primary_scale": "monumental", "secondary_scale": "small", "camera": "flat editorial orthographic", "cropping": "dramatic boundary intrusion", "negative_space": "dominant", "focal_point": "physical interaction" } def typography(self): return { "heading": derive_chapter_title(self.topic), "embedded_word": infer_single_action_label_if_needed( self.topic ), "density": "minimal" } def render_style(self): return { "background": "warm ivory", "foreground": "near-black", "interior_lines": "fine white contours", "shadows": None, "gradients": None, "texture": "minimal editorial print", "format": "wide conceptual illustration" } def render(self): return compose( actors=self.visual_actors(), metaphor=self. best_metaphor, layout=self.composition(), typography=self.typography(), style=self.render_style() ) FINAL_IMAGE = UniversalEditorialMetaphor( topic="[TOPIC]" ).render()
This image was generated with Nano Banana 2.1 by Google. The prompt on this page is the exact text used to create it.
The example was generated with Nano Banana 2.1 using: 120:67 · Poster · Cinematic · Sci-Fi. Reusing these settings gives the closest match.
Copy the prompt as-is first, then replace only the subject while keeping the style, lighting, and composition keywords. Generate a few variations — outputs differ between runs.
Yes. Prompts work best as starting points — adjust descriptors, aspect ratio, or style tags to make the output your own.