Welcome to my 💫🌄 Friendly Krea 2

✨ Less mess, more magic

Make all your fantasies come true with Krea 2!

I offer my workflow with Super Resolution 6MP mode, 3 modes LLM Enhancer and other optimizations. Supported T2I and I2I image reference generation. Easy control of each part of your generation!

The latest version of ComfyUI is required.

💻 System requirements:

• Minimum system requirements:

RTX 3000-s, 8GB+ VRAM, 32GB+ RAM, 8-core processor, SSD, latest ComfyUI

📌 Detailed tips and links to models in the workflow

✨ Workflow features:

• Extremely user-friendly interface

• Maximum performance even on low-end systems with at least 8GB of VRAM

• Optimized LLM Enhancer implementation

• Support for up to 4 Lora

• Detailed tooltips and links to models

• Manual random seed for complete control over generations

Model overview

Krea 2 is an advanced, high-performance AI image generation model developed by Krea AI. Designed for both speed and exceptional visual fidelity, it represents a major shift from traditional tag-based generation to deep natural language understanding.

Key Highlights:

• Natural Language Understanding: Powered by a modern text encoder (Qwen3-VL), it excels at interpreting fluid, descriptive paragraphs and novelistic storytelling rather than relying on comma-separated keyword tags.

• High Resolution: Capable of generating highly detailed, cinematic images up to 2K resolution (particularly in its Turbo variant).

• Precise Text Rendering: Accurately renders specific typography on signs, banners, or clothing when the exact words are enclosed in double quotation marks.

• Complex Composition Control: Handles long, comprehensive prompts exceptionally well, allowing for precise control over lighting, camera angles, spatial positioning, and material textures.

• Speed & Workflow Integration: Built on Krea's legacy of fast-generation workflows, it is highly optimized for rapid iteration and integrates seamlessly into advanced pipelines like ComfyUI.

🤗🙏🏼 Thanks to Krea AI

Original repo — GitHub