Deepseek&Janus: Img generation and understanding

Introduction:

ComfyUI now includes the DeepSeek Janus-Pro Models: Janus-Pro-1BJanus-Pro-7B, This workflow offers flexibility in image creation while ensuring high-quality image analysis. It’s ideal for users needing both AI-driven visual generation and intelligent scene interpretation.

Workflow Overview:

How to use this workflow:

1. Image Generation

This workflow allows users to generate images based on text prompts using Janus Image Generation. Users can switch between two models for different performance and quality needs:

• Janus-Pro-1B → Faster, lightweight, suitable for quick image generation.
• Janus-Pro-7B → More detailed, higher-quality results, better for complex prompts.

Process:

1. Model Selection:Users can choose between Janus-Pro-1B and Janus-Pro-7B via the Janus Model Loader node.

1. Prompt Input:
2. Output:The generated image is displayed using the Preview Image node.

2. Image Understanding

This part of the workflow analyzes an uploaded image and generates a detailed textual description.

Process:

1. Image Upload:Users can upload an image via the Load Image node.
2. Text Output:The Show Text node displays the generated description.