Description: 🚀 Run this workflow for FREE on Runninghub:
https://www.runninghub.ai/post/2072151051746242562/?inviteCode=rh-v1497
(New users get 1,000 free credits + 100 daily login points—grab it while the promo lasts!)
🎥 Watch the full YouTube breakdown:
💬 Join Discord for the Prompt Templates:
🌟 Overview Welcome to the 12-billion-parameter beast: Krea 2. While the raw version of Krea 2 is built for developers and AI "alchemists," this ComfyUI workflow makes its Turbo variant accessible to everyone. To solve the issue of annoying online LLM node installations, I have completely integrated a local Qwen 3.5 model to handle your prompt expansion automatically!
⚙️ Core Features:
Local LLM Prompt Structuring: Krea 2 requires highly structured prompts to shine. This workflow uses a built-in Qwen 3.5 node and a custom template to take your simple ideas and automatically expand them into Krea-friendly English prompts locally.
3-Pass Sampling Architecture: Ensures high-quality generation with precise step control.
Klein Upscaler Integration: Directly hooks into a Klein upscaler at the end of the pipeline for crisp, high-definition final outputs.
💡 Usage Tip: Just set your dimensions, type a simple prompt in the designated node, and let the Qwen LLM do the heavy lifting.