PixOnDeck — Image to Prompt

Turn images into editable prompts with PixOnDeck credits or your own API key.

Turn images into editable English image-generation prompts in Chrome's side panel. • Select a webpage image or upload one from your computer. • Edit and copy prompts, or continue creating in ChatGPT or PixOnDeck. • Clean up explicit descriptions and switch between the original and cleaned prompts. • Recognize poses locally after a one-time model download. Preview skeletons with or without the source image, and export PNG or JSON. • Optionally recognize poses after image-to-prompt tasks and include a skeleton reference in PixOnDeck. ChatGPT requires manual pasting of the copied reference image. • Revisit recent tasks for seven days and save up to 100 prompts to your private library, shared with the PixOnDeck website. A PixOnDeck account is required. Image-to-prompt and prompt-cleaning tasks each cost 1 PixOnDeck credit when completed. You can instead connect your own API key through an OpenAI-compatible HTTPS endpoint. In BYOK mode, selected images or text go directly to your provider, which handles billing; no PixOnDeck credits are used, and your key stays in browser-session storage. Local pose recognition uses no credits. Website image generation is charged separately. Only images you select are submitted for image-to-prompt tasks. Pose recognition runs locally; sending a skeleton reference to PixOnDeck is optional. Prompts and pose estimates may need editing. Prompt cleaning does not guarantee acceptance by any platform. Some websites restrict image access; local upload is available as an alternative. Available in English and Simplified Chinese.