WorkflowListedExternal

Flux.2-Klein Image To Image editing — v2.0 — v4.0

VersionRelease 3270051 — flux2KleinImageTo_v40.jsonEstimated VRAMNot yet characterized

Runtime profile

Access: External workflow
Status: Listed
Version: flux2KleinImageTo_v40.json
Source: Civitai
Updated: Oct 15, 2026

Source description

NEW MODE T2I or I2I with IDENTITY GUIDANCE This method replaces the PulId system in a simpler and more efficient way—without the need for a LoRA, simply by using the FLUX.2 Klein Identity Guidance node; It is now very simple to maintain identity during editing and new image generations, see the inside Markdown.

What you need

Model files: 7.7 GiB known subtotal

Remaining hardware or download-size estimates are not available yet.

AI-generated commentary

AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.

Purpose

Use this ComfyUI workflow with FLUX.2 Klein to generate images from text prompts or edit reference images; it offers identity guidance and produces a generated or edited image.

Flow

The workflow includes image-loading paths for inpainting, face cropping, outpainting, and background removal, along with image-editing and identity-guidance nodes.

Inputs and outputs

For image editing, provide a prompt and one or more reference images; version 4 describes choosing new dimensions for the resulting image.
The workflow includes nodes to preview and save generated images.

Setup

The author reports using ComfyUI with a 16 GB VRAM card and says missing nodes should install without major issues.

Dependencies

The listed ComfyUI node packs are comfyui-crystools, comfyui-easy-use, comfyui-flux2klein-enhancer, comfyui-imgnr-utils, comfyui-n-nodes, comfyui-rmbg, facerestore_cf, gguf, and rgthree-comfy.
Listed model files include Flux2klein/flux-2-klein-9b-turbo-fp8.safetensors, klein_snofs_v1_4.safetensors, and Qwen3_8b_abliterated_v2-fp8mixed.safetensors.
Other listed model files are RealESRGAN_x2plus.pth and vae_flux.2-klein_diffusion_pytorch_model.safetensors.

Configuration

Version 4 adds controls for choosing output dimensions when editing an image; the included Markdown notes cover the details.
You can bypass the included upscaling step when you do not want to use it.
Suggestion · not verified

Cautions

The author warns that prompt weights such as (word:weight) may cause visual artifacts, distorted colors, or missing elements.
The author says facial resemblance can be limited with this workflow.

Does this need to be edited?

Sources

1 source

CivitaiPrimaryVersion 3270051

civitai.com/models/2775812

Verified Oct 14

Source excerpts

1 excerpt

NoteOct 14civitai.com/models/2775812

Source context: 1060 downloads · Type Workflows · Base model Flux.2 Klein 9B-base