<h1 id="flux.2-klein-9b-ultimate-all-in-one-comfyui-workflow-v2.0">FLUX.2 Klein 9B — Ultimate All-in-One ComfyUI Workflow v2.0</h1>90 nodes · 6 pipelines · 12 organized groups · 8GB VRAM compatibleOne workflow file. Everything you need. No more switching between different JSON files for different tasks.<hr /><h2 id="6-pipelines-in-1-workflow">6 Pipelines in 1 Workflow</h2>Pipeline: What It Does... 🎨 F1: Text → ImageStandard txt2img at 1024×1024 🖼️ F2: Single-Reference KV EditEdit specific elements while preserving everything else 🚀 F3: Face + Pose SwapTake a face from Image A + pose from Image B → combine realistically 🎭 F4: InpaintingManual mask or AI auto-masking → regenerate only masked area 🔀 F5: Image MergeBlend two images with adjustable ratio ✨ RefinerDetail enhancement, lighting correction, skin texture improvement<h2 id="built-in-features">Built-in Features</h2><ul><li>🧭 NAG (Normalized Attention Guidance) — Negative prompts that actually work on distilled Flux</li><li>🤖 Florence2 Auto-Masking — Type what to mask ("Segment the shirt") → pixel-perfect mask</li><li>⬆️ 4x UltraSharp Upscaler — Clean upscaling with detail preservation</li><li>🔷 Tiled VAE Decode — All VAE decodes are tiled to prevent OOM on 8GB VRAM</li><li>🔗 2-Slot LoRA Chain — Enhancer LoRA always last, add your own LoRAs in the empty slot</li><li>📋 Built-in Guide Notes — Download links, prompting tips, and troubleshooting right inside the workflow</li></ul><h2 id="hardware-requirements">Hardware Requirements</h2><ul><li>Minimum: 8GB VRAM (RTX 3060/4060/4070 with GGUF Q4 model)</li><li>Recommended: 10-12GB VRAM (RTX 3080/4070 Ti with FP8 model)</li><li>RAM: 16GB+</li><li>ComfyUI: Version 0.9.2 or newer</li></ul><h2 id="models-you-need">Models You Need</h2>Diffusion Model (pick one):<ul><li>FP8: <code>flux-2-klein-9b-fp8.safetensors</code> → <a target="_blank" rel="ugc" href="https://huggingface.co/black-forest-labs/FLUX.2-klein-9b-fp8">Download</a></li><li>GGUF Q4 (8GB VRAM): → <a target="_blank" rel="ugc" href="https://huggingface.co/unsloth/FLUX.2-klein-9B-GGUF">Download</a></li></ul>Text Encoder:<ul><li><code>qwen_3_8b_fp8mixed.safetensors</code> → <a target="_blank" rel="ugc" href="https://huggingface.co/black-forest-labs/FLUX.2-klein-9B">Download</a></li></ul>VAE:<ul><li><code>flux2-vae.safetensors</code> → <a target="_blank" rel="ugc" href="https://huggingface.co/black-forest-labs/FLUX.2-klein-9B">Download</a></li></ul>Upscale Model:<ul><li><code>4x-UltraSharpV2.pth</code> → <a target="_blank" rel="ugc" href="https://huggingface.co/Kim2091/UltraSharpV2/tree/main">Download</a></li></ul>LoRA:<ul><li><code>klein_9b_enhancer_v2.safetensors</code> → Put in <code>models/loras/</code> <a target="_blank" rel="ugc" href="https://civarchive.com/models/2442399/elusarcas-detail-enhancer-or-flux-klein-9b?modelVersionId=2746136">download</a></li></ul>Custom Nodes (install via ComfyUI Manager):<ul><li>rgthree-comfy (Image Comparer)</li><li>ComfyUI-Florence2 by kijai (auto-masking, optional)</li><li>ComfyUI-GGUF by city96 (only if using GGUF models)</li></ul><h2 id="how-to-use">How to Use</h2><ol><li>Load the JSON in ComfyUI</li><li>use comphyui manager to install any missing node. ( critical step )</li><li>Only the Refiner is active by default — everything else is bypassed</li><li>To activate a pipeline: right-click its group header → Set Nodes Mode → Always Execute</li><li>To deactivate: right-click → Set Nodes Mode → Bypass ( or bypass groupe nodes )</li><li>Read the built-in Note nodes for prompting tips and download links</li></ol><h2 id="prompting-tips">Prompting Tips</h2>FLUX.2 Klein reads prompts like a novelist, NOT like Stable Diffusion.Use: <code>Subject → Setting → Details → Lighting → Technical terms</code>Power words: <code>clean digital file</code>, <code>histogram equalization</code>, <code>gamma correction</code>, <code>studio strobe</code>, <code>shot on Canon EOS R5</code>Avoid: <code>beautiful</code>, <code>masterpiece</code>, <code>best quality</code>, <code>trending on artstation</code>For text in images: Put text in quotes → <code>A sign reading "HELLO"</code><h2 id="troubleshooting">Troubleshooting</h2><ul><li>OOM: Already using Tiled VAE. Try GGUF Q4 model if still crashing</li><li>Black images: Wrong text encoder. 9B needs Qwen3 8B, not 4B</li><li>Plastic faces: Make sure enhancer LoRA is active (strength 0.7)</li><li>NAG not working: Install ComfyUI-NAG custom node as alternative</li></ul><hr />Built and tested on RTX 4060 Mobile (8GB VRAM). If you have issues, leave a comment — I'll help you fix it.If this workflow is useful to you, please leave a review — it helps others find it.<h2 id="version-history:">VERSION HISTORY:</h2><ul><li>v2.1 — Initial public release. +90 nodes, 6 pipelines, NAG, Florence2, Tiled VAE. added groupe bypasser, notes for new people to comphyui</li></ul>

Files with SHA256: bd7fb0754ffb1d03fa751b881f7ed72ae3191dca7f115add846bf8edec08c8c8