<h2 id="qwen-image-edit-2511-2-image-edit-workflow-(base-+-reference)">Qwen Image Edit 2511 — 2-Image Edit Workflow (Base + Reference)</h2>This workflow uses Qwen Image Edit 2511 to perform two-image guided editing: you provide a Base Image (the image you want to edit) and a Reference Image (the image that guides style, clothing, identity cues, composition, or the specific edit you want). The model blends your prompt with the reference guidance to produce controlled edits while keeping the base image structure intact. <h3 id="worflow-tested-on-gpu:-nvidia-rtx-4060ti-16gbram:-64gb-ddr5generation-time:-90-150-seconds-(4-6-steps)">Worflow Tested On GPU: Nvidia RTX 4060ti 16GB RAM: 64GB DDR5 Generation Time: 90-150 seconds (4-6 steps) </h3><h3 id="what-this-workflow-is-for">What this workflow is for</h3><ul><li>Outfit / style swap using a reference image</li><li>Background replacement guided by reference</li><li>Face/detail enhancement (with deblur LoRA)</li><li>Product/mockup edits with consistent lighting/style</li><li>“Make my image look like this” edits without completely re-generating from scratch</li></ul><h3 id="inputs-(2-images)">Inputs (2 Images)</h3><ol><li>Base Image (Required) The original image you want to modify.</li><li>Reference Image (Required) The guiding image: style, clothing, pose cues, composition, material details, or “target look”.</li></ol><h3 id="how-it-works-(high-level)">How it works (high level)</h3><ul><li>The workflow loads:<ul><li>Qwen Image Edit 2511 diffusion model</li><li>Qwen 2.5 VL text encoder</li><li>CLIP Vision for image understanding</li><li>Qwen Image VAE</li></ul></li><li>Both images are encoded and used as conditioning:<ul><li>Base image anchors structure and layout.</li><li>Reference image drives the requested changes (style/appearance/details).</li></ul></li><li>Prompt controls what to keep vs what to change:<ul><li>Use “preserve” language to keep base details.</li><li>Use “apply from reference” language to pull details from the reference.</li></ul></li></ul><h3 id="prompting-tips-(works-insanely-well)">Prompting tips (works insanely well)</h3>Structure:<ul><li>Keep / preserve: what must stay from base image</li><li>Apply from reference: what to transfer</li><li>Constraints: realism, lighting, material, no extra changes</li></ul>Example prompt:Edit the base image. Preserve face identity, skin texture, and body proportions from the base image. Apply hairstyle, outfit design, and accessories from the reference image. Keep the same camera angle and pose as base. Photorealistic materials, clean lighting, high detail, no extra accessories.Negative prompt ideas:blurry, low quality, extra fingers, warped face, deformed, wrong outfit details, inconsistent lighting, text, watermark, logo<h3 id="recommended-starting-settings">Recommended starting settings</h3><ul><li>Steps: 4</li><li>CFG: 1</li><li>Denoise/Strength: 1</li></ul><h3 id="lora-usage-(optional-but-recommended)">LoRA usage (optional but recommended)</h3><ul><li>Lightning 4-Steps LoRA Great for faster edits and stable results at low steps.</li><li>Deblur LoRA Use when details look soft, especially faces/textures. Keep LoRA strength modest to avoid over-sharpen artifacts.</li></ul><h3 id="best-practices">Best practices</h3><ul><li>Use a clean, high-res reference image with clear outfit/details.</li><li>If the model changes too much of the base image: reduce denoise.</li><li>If the model ignores the reference: increase denoise slightly and clarify “apply from reference”.</li><li>For consistent results, keep lighting constraints in the prompt (e.g., “same lighting as base”).</li></ul><h3 id="output">Output</h3>Produces an edited image that keeps the base composition while applying changes guided by the reference image + prompt.

Files with SHA256: 9fd58ce2f79a22abb5c504dc925839e6bb63d6c15ecd90d90b8280fe18acafd0