A ComfyUI node uses local AI models to make image prompts
Llama Prompt Generator is a single ComfyUI node that runs a local LLM inside the node for prompt work. It can use or Ollama as the backend. For text, it can improve a prompt using a chosen .
For images, it can analyze a loaded image and turn it into a prompt or description. It lets people save favorite as presets and switch between them quickly. It also supports refining generated prompts, comparing changes, and viewing version history.
Model settings can be handled inside the node, including backend, model, vision options, and some Ollama actions such as starting, stopping, and pulling models. It also shows live token streaming and can run the model from a Generate button without queueing the full ComfyUI workflow unless the node output is used elsewhere.
Key points
- The node runs a local LLM inside ComfyUI for prompt-related work.
- It supports text prompt improvement and image-to-prompt analysis.
- Favorite can be saved as presets and swapped quickly.
- Generated prompts can be refined, compared, and tracked through version history.
- Ollama users can manage some model actions from inside the node.