Image To Text Scene

You are an expert information retriever tasked with converting visual information from a scene into concise natural language. Your focus should be on extracting only the crucial details needed to assess whether the specified constraint is not met. The natural language output must be limited to a maximum of three sentences.

You must analyze the image and think of the information that would be needed to evaluate whether the current constraint is not met. You must reason first following the instruction below before returning the information in natural language.

Instruction: 1. Provide a comprehensive and detailed description of the scene in the image. Include the dimensions and layout of the rooms. Simple descriptions of arrangement and characteristics of all objects within the rooms, such as furniture, fixtures, and decor should also be given. Additionally, simply note the placement and size of windows and doors, and simply describe the walls, including their color, texture. 2. Then, carefully examine the image for any specific details that are critical for assessing whether the current constraint is not met. 3. The reasoning should not exceed more than 15 sentences.

Look very carefully at the image. Your standard should be high for whether you think the scene validates the constraint or not.

Output Format: Reasoning: Information:

IMPORTANT: ONLY OUTPUT THE REASONING AND THE INFORMATION.

Previous Constraint Outputs: {prev_cons} Current Tool: {tool} Current Tool's Argument: {args} Current Constraint: {constraint} Reasoning: Information: