<p><em>Part 3 of my Lab series. Checkout my other workflows on my profile, they can all be used in conjunction!</em></p><p><em>Sorry for the clutter! - subgraphs seem to break during upload</em></p><p><em>Unpacked - removed all subgraphs since they glitched sometimes. Everything displayed on 1st layer canvas</em></p><h1 id="swaplab-flux-2-klein">⚡ swapLab Flux 2 Klein</h1><h3 id="the-ultimate-dual-image-characteristic-transfer-workflow-for-comfyui"><em>The Ultimate Dual-Image Characteristic Transfer Workflow for ComfyUI</em></h3><p>Load two images. Pick what to transfer. Watch the magic happen.</p><p>SwapForge is a <strong>ComfyUI workflow</strong> that takes any two photos and seamlessly transfers faces, hair, clothes, and more from one person onto another — producing composites so clean they look like they were shot that way. No Photoshop. No manual masking. No fuss.</p><hr /><h2 id="why-swaplab-is-so-effective">🚀 Why swapLab is so effective</h2><h3 id="powered-by-flux-2-klein-9b-fast-doesn't-mean-cheap">⚡ Powered by Flux 2 Klein 9B — Fast Doesn't Mean Cheap</h3><p>Forget waiting minutes for mediocre results. swapLab runs on <strong>Flux 2 Klein 9B</strong> — one of the most advanced open-source image editing models on the planet. Klein's reference-guided architecture was <em>built</em> for exactly this kind of task, delivering <strong>cinema-quality swaps at speeds that will make your jaw drop</strong>. Iterate faster. Create more. Wait less.</p><h3 id="any-image.-any-size.-zero-prep.">📐 Any Image. Any Size. Zero Prep.</h3><p>Got images from different cameras, different aspect ratios, different resolutions? <strong>Don't touch them.</strong> swapLab's intelligent autoscaling system reads your input dimensions and automatically calibrates every part of the pipeline — latent canvas, preprocessor resolution, generation size — all of it. Just drop your images in and go.</p><h3 id="you-stay-in-control">🎯 You Stay In Control</h3><p>swapLab works beautifully with zero prompt input — but when you <em>want</em> to get specific, the guided prompting system lets you steer the output with natural language. Want to preserve a particular detail? Emphasize something? Tweak the vibe? <strong>Just tell it.</strong> Your creative vision, turbocharged by AI.</p><h3 id="reactor-face-lock-identity-that-actually-sticks">🔒 ReActor Face Lock — Identity That Actually Sticks</h3><p>Generative models drift. Faces shift. Not here. After every swap, swapLab runs a <strong>ReActor face reinforcement pass</strong> using your original source image — mathematically anchoring the final face to your reference identity. The person in the output <em>is</em> the person in your photo. No blending artifacts. No uncanny valley. Just a face that holds.</p><hr /><h2 id="swap-modes">🎛️ Swap Modes</h2><p>Mode What Transfers</p><p>👤 <strong>Face &amp; Hair </strong>Facial features, expression, hairstyle</p><p>👗 <strong>Outfit</strong>Full clothing transfer, accessories</p><p>🧍 <strong>Full Body </strong>Complete person into new context</p><p>✨ <strong>Custom</strong> You pick — any combo of face, hair, body, clothes, accessories, background</p><hr /><h2 id="under-the-hood">🛠️ Under the Hood</h2><ul><li><p>🧠 <strong>Flux 2 Klein 9B</strong> — reference-guided generation engine</p></li><li><p>🎭 <strong>PersonMaskUltra V2 + SAM3</strong> — surgical region detection and masking</p></li><li><p>📏 <strong>Intelligent Autoscaling</strong> — universal image compatibility</p></li><li><p>💬 <strong>Guided Prompting</strong> — natural language creative control</p></li><li><p>🔐 <strong>ReActor</strong> — identity-locking face reinforcement pass</p></li></ul><hr /><p></p>

Files with SHA256: aa368cece8106110ca55a5165f3153e39be22e61d5d1b0bee6eef1196cebcf16