---
title: "Build a ComfyUI Image Edit Workflow and Actually Understand It | Build With Me [EP. 4]"
source_url: https://youtu.be/k1STt31lsDA
captured_at: 2026-07-11
author: "LoRAtech - ComfyUI Hub"
contributor: "ชิดชนก"
platform: youtube
---
# Build a ComfyUI Image Edit Workflow and Actually Understand It | Build With Me [EP. 4]

youtube video by **LoRAtech - ComfyUI Hub**, 18m54s, uploaded 2026-07-08. URL: https://youtu.be/k1STt31lsDA

## Shared in LINE Note+
- 2026-07-11 00:51 by ชิดชนก

## Description
Blank canvas to a working Flux 2 Klein image edit workflow — 17 nodes, every single one explained. This workflow takes two reference photos (your character + a product, outfit, or scene) and combines them into one new image with a single prompt. No masks, no inpainting, no LoRA training. By the end you won't just have the workflow, you'll understand why every node is there.

🧠 WHAT YOU'LL LEARN
▶ How Flux 2 Klein edit workflows actually work (references live in the conditioning, not the latent)
▶ Chaining ReferenceLatent nodes for multi-image editing in ComfyUI
▶ Why reference images get scaled to 1.2 megapixels before encoding
▶ Prompting for edits: directing the model instead of describing a scene
▶ Character consistency from a single photo — no dataset, no training
▶ Fixing reference takeover when the second image starts hijacking your subject

🔗 FREE WORKFLOWS, COMMUNITY DISCUSSION, 1 ON 1 SESSIONS
If you want 1-on-1 coaching for your ComfyUI/RunPod setup, LoRA training, or growing an AI influencer account, sessions are consultation-based — details inside: 

⚡ RUN COMFYUI ON RUNPOD
 — add a minimum of $10 through the link and get $5 free

📺 THE BUILD WITH ME SERIES
EP. 1 — Buil