T8mars/Comfyui-zhenzhen

What it solves

This project provides a comprehensive set of custom nodes and workflows for ComfyUI that allow users to access a wide variety of high-end AI models (image, video, audio, and 3D) via APIs. It simplifies the process of integrating expensive or complex frontier models—such as Sora, Midjourney, Grok, and Gemini—into a visual node-based workflow without requiring the user to host these massive models locally.

How it works

The repository implements ComfyUI nodes that act as bridges to the "Zhenzhen AI Workshop" (and its domestic Chinese version) API services. Users configure an API key within the nodes to send requests to remote servers, which then process the generation tasks and return the resulting media (images, videos, or audio) back into the ComfyUI interface. It includes specialized logic for handling asynchronous tasks, polling for results, and managing media downloads from object storage (OSS).

Who it’s for

• ComfyUI users who want to use state-of-the-art AI models (like Sora 2, Midjourney v8, or Gemini 3.1) without owning high-end hardware.
• Creators looking for a unified interface to experiment with multiple multimodal models (text-to-video, image-to-video, audio generation) in a single workflow.
• Users in China seeking affordable access to overseas AI models through a specialized API proxy.

Highlights

• Extensive Model Support: Integrates a vast array of models including Sora 2, Midjourney, Grok Video, Gemini, Suno, and various Minimax/Kling/Vidu models.
• Multimodal Capabilities: Supports text-to-image, image-to-video, image-to-3D, audio-to-speech, and music generation.
• Advanced Workflow Features: Includes support for concurrent generation (up to 30 images and 10 videos), multi-reference image inputs, and custom resolution/aspect ratio controls.
• Ready-to-use Workflows: Provides numerous pre-configured JSON workflow files for specific tasks like character consistency, video editing, and high-resolution upscaling.