skills.rest
artokun
artokunCOMMUNITY
Explore all skills by this author →
Website𝕏@artokunGitHub

krea2-txt2img

Build Krea 2 Turbo txt2img workflows in ComfyUI with Qwen3-VL encoding and JSON prompting.

715|111|Updated Feb 15, 2026
Search Tags:#image generation#Workflow#diffusion#ComfyUI#txt2img#JSON prompting#krea 2
One-click install
npx skills add https://github.com/artokun/comfyui-mcp --skill krea2-txt2img
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: krea2-txt2img
Source: https://github.com/artokun/comfyui-mcp/tree/main/plugin/skills/krea2-txt2img
Command: npx skills add https://github.com/artokun/comfyui-mcp --skill krea2-txt2img

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Setting up a working Krea 2 Turbo text-to-image pipeline in ComfyUI requires knowing the correct CLIPLoader type, text encoder, VAE, sampler settings, and node wiring, which is easy to get wrong without a verified reference.

Core Features & Use Cases

  • Three ready packs: choose manual prose prompting, Ideogram-4-style structured JSON prompting, or a two-pass detail-boost combo with turbo LoRA.
  • Verified turbo settings: 8 steps, cfg 1, er_sde sampler, simple scheduler, 1920x1080 default, with the Krea2T-Enhancer detail patch active.
  • Structured JSON prompting guidance: configure the Ideogram4PromptBuilderKJ node with background, technical, style, and lighting fields while avoiding palette and stale-field pitfalls.
  • Use Case: You want a crisp 1920x1080 product still-life from a structured JSON prompt; load the krea2-txt2img-json pack, set all builder fields, render in 8 steps, and verify the output matches the JSON.

Quick Start

Ask the agent to build a Krea 2 Turbo txt2img workflow using the krea2-txt2img-json pack with a structured JSON prompt for your scene.

Frequently Asked Questions about krea2-txt2img

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a Krea 2 txt2img workflow in ComfyUI?▼

Use one of the three packs: krea2-txt2img-manual for prose prompts, krea2-txt2img-json for structured JSON prompting, or krea2-combo for a two-pass detail boost. Each wires UNETLoader, CLIPLoader with type krea2, and the Qwen image VAE into a KSampler subgraph.

What sampler settings work best for Krea 2 Turbo?▼

Krea 2 Turbo is distilled, so use 8 steps at cfg 1 with the er_sde sampler and simple scheduler. More steps or higher cfg over-cooks the output, and 1920x1080 is the verified default resolution.

Why does CLIPLoader say krea2 is not in the list?▼

This error means your ComfyUI version is too old for native Krea 2 support. Update ComfyUI to v0.26.0 or later, which includes the krea2 text encoder type in comfy/text_encoders/krea2.py.

How do I use JSON or area prompting with Krea 2?▼

Use the krea2-txt2img-json pack, whose Ideogram4PromptBuilderKJ node feeds structured prompts to the Qwen3-VL encoder. Set all builder fields including background, technical, style, and lighting, and keep palettes minimal or empty to avoid color-swatch artifacts.

What models does Krea 2 Turbo require in ComfyUI?▼

You need the krea2_turbo fp8 diffusion checkpoint, the qwen3vl_4b_fp8_scaled text encoder, and the qwen_image_vae. The combo pack additionally uses the krea2 turbo LoRA at 0.2 strength on both passes, with IdeoKrea as an optional add-in.

Related Skills

Explore More Design & Creative Skills→
jackspacejackspace
16

comfyui-workflow-helper

Guide ComfyUI workflow creation, node configuration, and optimization.

Community
Advanced
zkzupupzkzupup

excalidraw-workflow

Export Excalidraw diagrams to PNG and validate JSON rendering rules.

Community
Intermediate
marmikcfcmarmikcfc

aeo-visibility-monitor

Track AEO visibility score changes and alert on significant shifts.

Community
Advanced
doany-aidoany-ai
5

image-edit

Route image edits to the best RunComfy model and download results.

Official
Intermediate