👁️ Vision AI • 2D Bounding Boxes • Chart & OCR Extraction • High-Resolution Tiling
Multimodal Vision Prompt Compiler
Generic questions like “what is in this image?” fail on small details. Structure production vision prompts with normalized 2D bounding boxes, structured markdown table schemas, and spatial coordinate grids.
Vision Task Parameters
💡 Image Tile Token Cost
A 1080p image in high-detail mode is split into 4 tiles of 512x512 pixels + 1 low-res thumbnail = ~765 input tokens ($0.002).
Compiled Vision Prompt & JSON Schema