ComfyUI_Prompt-All-In-One

中文|English

ComfyUI Nodes for Generating Prompts for All Video, Audio, Image, and Text Creations

Added support for all Ollama models. Any Ollama model can run, including custom models. Locally generate any prompt, reverse-prompt images, videos, and more. Have fun!

Added audio/music reverse-prompting and tagging, a powerful tool for audio/music LoRA training. It's fast, effective, and even surpasses API performance.

Updated JoyCaption to the latest version, with NF4 support. A blessing for low VRAM users.

Currently supported APIs: DeepSeek-R1/V3, almost the entire Alibaba Cloud Qwen family, and the entire Google Gemini family. More useful APIs will be added successively.

Selected local models are provided separately for your convenience. Just download what you need.

📣 Updates

[2025-06-12]⚒️: Released version v2.0.0.

[2025-05-10]⚒️: Support for the latest Gemini model, gemini-2.5-pro-preview-05-06, which requires payment. Support for the latest Qwen3.

[2025-04-12]⚒️: Added support for JoyCaption.

[2025-04-01]⚒️: Released version v1.0.0.

• Supports DeepSeek-R1/V3 model API. You need to apply for an API key on the DeepSeek official website. Then, create a new environment variable DEEPSEEK_API_KEY = <your key> in your system's environment variables. For instructions, see Configure API Key through Environment Variables. Alternatively, you can skip using the system variable and enter the key directly into the node, but be careful to keep your key confidential and do not leak it with your workflow. On Windows, you may need to restart your computer for the environment variable to take effect.

• Supports Qwen API. Apply for an API key at Alibaba Cloud Bailian. Then, create a new environment variable DASHSCOPE_API_KEY = <your key> in your system's environment variables. The method for adding and using it is the same as above. Be careful to keep your key confidential and do not leak it with your workflow. On Windows, you may need to restart your computer for the environment variable to take effect. Inference models with the "_R" suffix have a thought process.

• Supports Gemini API. Apply for an API key at Google AI Studio. Then, create a new environment variable GOOGLE_API_KEY = <your key> in your system's environment variables. The method for adding and using it is the same as above. Be careful to keep your key confidential and do not leak it with your workflow. On Windows, you may need to restart your computer for the environment variable to take effect.

Usage

Audio/Music Reverse-Prompting:

Ollama Model Flux Prompt Generation:

Ollama Model Image/Video Reverse-Prompting:

JoyCaption Image Interrogation/Description:

• images_dir: Path for batch-tagging images.

• save_img_prompt_to_folder: Save path for images and prompts. If provided, batch-tagged images and their prompts will be saved to this folder. Otherwise, they are saved by default in images_dir with the same name as the image. If provided, even single images can be saved to this folder.

DeepSee