whatisgithub

What is comfyui-krea2-reference?

konokoaz/comfyui-krea2-reference — explained in plain English

Analysis updated 2026-05-18

17PythonAudience · vibe coderComplexity · 4/5Setup · hard

In one sentence

A ComfyUI add-on for the Krea 2 image model that combines a character reference and a style reference into one coherent AI-generated image instead of a collage.

Mindmap

A visual breakdown will appear here once this repo is fully enriched.

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Generate a new image of a specific character in a different art style using two reference images.

USE CASE 2

Combine a character reference and a text-described style to keep an existing character consistent across scenes.

USE CASE 3

Produce a multi-panel comic keeping the same character's look across each panel.

USE CASE 4

Restyle a character's colors by editing the reference image rather than the text prompt.

What is it built with?

PythonComfyUIKrea 2LoRA

How does it compare?

konokoaz/comfyui-krea2-reference0petru/sentimoagi-eval-official/forte
Stars171717
LanguagePythonPythonPython
Setup difficultyhardmoderatehard
Complexity4/53/54/5
Audiencevibe coderdeveloperdeveloper

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · hard Time to first run · 1h+

Requires ComfyUI plus separately downloaded Krea 2 model files, a specific text encoder variant, and a reference LoRA file.

The README does not state a license.

So what is it?

ComfyUI-Krea2-Reference is a custom add-on node for ComfyUI, a visual tool for building AI image generation workflows. It works with an image model called Krea 2, and it lets you feed in reference pictures so the model can copy a character's look from one image and a painting style from another, then combine them, similar to telling an AI "draw the character from picture one, in the art style of picture two." Normal image editing tools often just paste or blend reference images together, which can look collaged and unnatural. This node instead feeds the reference images into the model as part of its internal understanding of the prompt, the same way it was originally trained to read images, so the result looks like a single coherently generated picture rather than a copy-paste job. Paired with a companion add-on called a reference LoRA, a small file that adjusts the model's behavior, the tool follows your written scene description more closely instead of just recreating the reference images. The main node lets you supply a text prompt along with up to four reference images, labeling each one as either a character reference or a style reference directly in your prompt text. You can control how much visual detail each reference contributes, crop a reference to a specific area, and remove backgrounds automatically. A second node loads the reference LoRA file and lets you tune how strongly it influences the results, with a recommended default setting described as producing the most reliable character likeness. The README notes some current limitations: using a reference sheet with multiple views of the same character can accidentally produce multiple views in the output, and combining both a character reference and a style reference in one prompt only works partially for now. A future version is described as in progress to address these. To install it, you clone the repository into ComfyUI's custom nodes folder, then download the required model files and the reference LoRA separately.

Copy-paste prompts

Prompt 1
Help me install ComfyUI-Krea2-Reference and set up the required Krea 2 model files.
Prompt 2
Show me how to build a prompt that combines a character reference and a style reference with this node.
Prompt 3
Explain how the reference LoRA strength settings affect character identity versus scene detail.
Prompt 4
Help me troubleshoot why my two-reference prompt is only partially following the style image.

Frequently asked questions

What is comfyui-krea2-reference?

A ComfyUI add-on for the Krea 2 image model that combines a character reference and a style reference into one coherent AI-generated image instead of a collage.

What language is comfyui-krea2-reference written in?

Mainly Python. The stack also includes Python, ComfyUI, Krea 2.

What license does comfyui-krea2-reference use?

The README does not state a license.

How hard is comfyui-krea2-reference to set up?

Setup difficulty is rated hard, with roughly 1h+ to a first successful run.

Who is comfyui-krea2-reference for?

Mainly vibe coder.

Open on GitHub → Ask about another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.