{
"function": "generate_map_scene_with_visual_input",
"user_input": {
"location_text": "วัดพระแก้ว",
"input_image_analysis": "AI_VISION_ANALYSIS_OF_UPLOADED_IMAGE",
"target_language": "Thai"
},
"ai_logic_instructions": {
"step_1_identify_location_name": "Determine the 'map_label_text' based on priority:\n1. IF 'location_text' is provided -> Use it.\n2. ELSE IF 'input_image_analysis' indicates a famous landmark -> Use the official landmark name.\n3. ELSE (Generic/Unknown) -> Generate a short, descriptive noun phrase describing the place (e.g., 'Wooden Pier', 'Night Market', 'Mountain Road'). DO NOT invent a fake proper noun (e.g., do NOT write 'Siam Pier' if unsure).",
"step_2_analyze_visuals": "Extract visual details from 'input_image_analysis' (colors, textures, architecture) to guide the 'environment_context' and 'focus_subject'.",
"step_3_determine_surroundings": "Ensure the surrounding map texture matches the uploaded image's environment (e.g., if image shows water, map must imply waterways).",
"dynamic_variables": {
"final_location_name": "The determined name from Step 1, translated to 'target_language'.",
"visual_description": "A concise description of the environment derived strictly from the uploaded image or text location.",
"focus_vehicle_or_element": "An object seen in the image OR a contextually appropriate miniature (boat, car, person) if none exists.",
"typography_style": "Legible, clean font style suitable for the location type.",
"nearby_context_labels": "1-2 generic labels for nearby elements based on visual logic (e.g., 'River', 'Main Road', 'Entrance') to populate the map without hallucinating street names."
}
},
"final_image_prompt_template": "A surreal, hyper-realistic close-up scene of a miniature {focus_vehicle_or_element} positioned on a winding path that seamlessly transforms into a printed paper map of {final_location_name}. The path transitions naturally into the map’s graphics. The central focus is the clear, legible map label '{final_location_name}' written in {typography_style}. The surrounding area reflects the actual environment visible in the input image — {visual_description}. The map design is minimalist vector style, limiting text to the main label and simple generic markers like '{nearby_context_labels}' to ensure readability and avoid gibberish. Lighting and color palette match the mood of the provided image. Aspect ratio 1:1, high resolution, premium editorial quality."
}