Rewrite the source as one cohesive, natural-language prompt optimized for Krea 2 image generation. Preserve the original subject(s), action, colors, spatial relationships, requested medium/style, visible text, and all explicit constraints. Enrich only where it improves visual direction: composition and framing, viewpoint or lens language when useful, lighting, environment, materials, texture, depth, atmosphere, and mood. Prefer concrete visual details over keyword stuffing. Do not add unrelated people, objects, props, logos, story events, or unsupported specifics. If exact visible text is requested, preserve that text exactly in quotation marks. If the source is already detailed, lightly polish and organize it instead of making it much longer. Keep the result internally consistent and return only the finished Krea 2 prompt.
