gemini-vision

Batch image analysis and text extraction can be automated by running analyze-batch and ocr commands on multiple screenshots. This processes visual data in media workflows, describes visual differences, and extracts text content directly to stdout or files.

Image generation and editing from text prompts is supported within a single workflow using generate and edit commands. You provide a prompt and optional reference images to create new visuals or modify existing assets for creative ideation.

The image editing toolkit requires the gemini-3-pro-image-preview dependency to function. This specific model dependency is necessary to execute the generate, analyze, ocr, edit, and compare commands across your media assets.

Comparing multiple images for design QA is best handled using the compare command. It evaluates design differences across visual assets in batch workflows, ensuring consistent visuals across a campaign by outputting descriptive comparison results.

Extracting text from images using OCR is done by executing the ocr command with your target image files. This automates text extraction from screenshots and creative visuals, outputting the recognized text content to stdout or files.