Notes:

  1. API Key: Saved in your browser's local storage for convenience. It is not sent anywhere except to official Google API servers.
  2. API Usage: Shows an offline estimate of your usage against the free tier limits (RPM: Requests Per Minute, RPD: Requests Per Day). This tracker resets daily based on your local clock.
  3. Delay: Important for preventing API rate limit errors. You can review rate limits here. This setting is ignored unless "Process Sequentially" is active.
  4. Generation Type: Choose your output: a descriptive Caption, comma-separated Tags, or a Weight Brief for AI art prompts (e.g., Stable Diffusion).
  5. Precise Value: Controls the AI's creativity. 'Precise' provides factual, repeatable descriptions (low temperature). 'Creative' allows for more imaginative interpretations (high temperature).
  6. Processing Mode: By default, images are processed in fast, cost-effective batches. Select "Process Sequentially" for slower, one-by-one processing, which enables the "Delay" setting to help avoid rate limits on heavy usage.
  7. Trigger Words: Adds specific keywords to the beginning of every generated prompt.
  8. Prompt Enrichment: Adds extra instructions to the AI's main task, allowing you to guide its focus.
  9. Export as JSON & Format as Single Paragraph: These options are available only for the 'Caption' generation type. They format the output for technical use cases.
  10. Auto Rename File Pairs: Renames all output files into a sequential format (e.g., 1.png, 1.txt). Can be combined with Kohya_SS export for numbered training data.
  11. Export as Kohya_SS structure: Creates a ZIP file with the specific folder structure required for Kohya_SS LoRA training.

Options

Usage Instructions:

  1. Open Google AI Studio to create an API key.
  2. Paste the key into the "API Key" box.
  3. Input one or more images.
  4. Select your desired mode and model.
  5. Adjust optional settings if needed.
  6. Click "Start".