PirateDiffusion Guide
Overview
PirateDiffusion by Graydient AI is a powerful bot on Telegram that handles large language models and thousands of image models from various model families like FLUX, Pony, and SDXL. It's extremely fast and lightweight on mobile, allowing users to link complicated desktop workflows to a hashtag and use them on-the-go.
Features
- Pro image generation and custom bot conversations over chat
- Create 4k images in a few keystrokes without a GUI
- Use every major Stable Diffusion via chat
- Create themed bots like /animebot, which connects PollyGPT LLMs to StableDiffusion models
- Create unique workflows using loadouts, recipes (macros), widgets (visual GUI builder), and custom bots
What's New
- New LLM command: /llm
- Many FLUX checkpoints available
- Over 10,000 models available
- Custom bots can now retain memories
- Better render monitoring
- New bots available, including /xtralargebot, /animebot, and /lightningbot
- Plus plan members can now choose Llama3 70 billion parameter for their bots
- New recipes available, including #quick for fast HD images
Premade and Custom Bots
- BASICS: The LLM command
- Introducing "POLLY"
- Create your own LLM agent
- Introducing custom bots
- Activating your bot
What Makes a Good Prompt?
- A prompt is a complete description that tells the AI what image to generate
- Describe the whole picture, every time
- Use the word order to your advantage
- Length also matters
Positive and Negative Prompts
- Positive prompts contain the subject and supporting details of the image
- Negative prompts describe what you do not want in the image
- Use nested parenthesis to add more emphasis to certain words
- Use brackets to make negatives stronger
Pro Commands
- RENDER: Create images like a stable diffusion pro
- INLINE POSITIVES, NEGATIVES: Use positives and negatives to steer the AI
- TRANSLATE: Render images in over 50 languages
- RECIPE MACROS: Use prompt templates to save time
- COMPOSE: Create multi-prompt, multi-regional images
- BLEND (IP ADAPTERS): Fuse multiple images together
- AFTER DETAILER: Fix bad hands, eyes, and faces in your images automatically