<h1 id="flux.1-dev:-the-&quot;smart-render&quot;-pipeline-(v4.0)">🚀 Flux.1 Dev: The "Smart-Render" Pipeline (v4.0)</h1>Optimized for 16GB VRAM Cards (e.g., RTX 4080 Super) Current Version: v4.0 (The Architecture Overhaul)Stop wasting time rendering high-res images just to check a prompt. This workflow introduces a "Smart-Render Logic" that separates creative iteration from final production. It mimics a professional 3D studio workflow: Work in Proxy, Render in Final.<hr /><h2 id="what's-new-in-v4.0">🆕 WHAT'S NEW IN v4.0?</h2><ul><li>⚡ 50% Draft Scaling: The Draft Mode now automatically scales resolution to 50%. This makes prompt testing lightning fast.</li><li>🖐️ 5-Stage Detailer Stack: Expanded from 3 to 5 slots. Now includes dedicated slots for Face, Hand, Person, and two slots for Anatomical/Feature enhancement (perfect for NSFW or highly detailed character work).</li><li>🎞️ The Analog Finish: A new post-processing chain combining Sharpening (before grain) and Film Grain (anti-banding) for a non-plastic, photorealistic look.</li><li>👁️ Visual Dev Station: A 3-monitor comparison setup inside the workflow to check "Raw vs. Detailer" and "Detailer vs. Grain" instantly.</li></ul><hr /><h2 id="the-logic:-the-&quot;red-button&quot;">💡 THE LOGIC: The "Red Button"</h2>I designed this workflow around a single Boolean Switch (The Big Red Button). It completely reconfigures the internal pipeline based on your goal.<h3 id="false-draft-mode-(speed-focus)">🔴 FALSE = DRAFT MODE (Speed Focus)</h3><ul><li>Resolution: Automatically scaled to 50% (via Aspect Ratio).</li><li>Model: Loads the fast Q4 GGUF (Quantum 4).</li><li>Pipeline: Bypasses Upscalers, Detailers, and Grain.</li><li>Save Path: Sorts images into a <code>..._LowRes</code> folder.</li><li>Goal: Instant iteration. Check your composition, lighting, and seed in seconds.</li></ul><h3 id="true-final-mode-(quality-focus)">🟢 TRUE = FINAL MODE (Quality Focus)</h3><ul><li>Resolution: Full 100% Output.</li><li>Model: Loads the heavy Q8 GGUF (Quantum 8) for maximum fidelity.</li><li>Pipeline: Activates the full chain: Upscale (x2) -&gt; 5x Detailer Stack -&gt; Analog Finish.</li><li>Save Path: Sorts images into a <code>..._Final</code> folder.</li><li>Goal: Production-ready export for Civitai, ArtStation, or Print.</li></ul>ℹ️ Why this works: As proven in testing, the Q4 and Q8 models are structurally almost identical. You can compose in Draft (Q4) and trust that the Final (Q8) render will keep the exact same pose and expression, just with much higher texture fidelity.<hr /><h2 id="key-features">🛠 KEY FEATURES</h2><ul><li>Hybrid GGUF Loading: Automatically swaps models in RAM to save VRAM headroom.</li><li>Non-Blocking Bus: You can turn off individual Detailers (e.g., disable Hand Detailer) without breaking the chain.</li><li>Smart Folder System: Automatically names files based on your <code>Project</code> and <code>Topic</code> inputs.</li><li>16GB VRAM Optimized: Runs smoothly on 4080-class cards by managing memory efficiently between the Upscaler and the Detailer Stack.</li></ul><hr /><h2 id="requirements-(download-checklist)">📦 REQUIREMENTS (Download Checklist)</h2>Please read the Notes inside the workflow!<ol><li>Checkpoints: You need TWO GGUF files of your preferred Flux model (e.g., CyberRealistic Flux).<ul><li><code>*Q4_k_s.gguf</code> (for Draft)</li><li><code>*Q8_0.gguf</code> (for Final)</li></ul></li><li>CLIP/VAE: Standard Flux files (<code>t5xxl_fp8</code>, <code>clip_l</code>, <code>ae</code>).</li><li>Detailer Models: <code>face_yolov8m.pt</code>, <code>hand_yolov8s.pt</code>, <code>person_yolov8m-seg.pt</code> (placed in <code>/models/ultralytics</code>).</li><li>Custom Nodes: If you see red nodes, simply use ComfyUI Manager -&gt; Install Missing Custom Nodes.</li></ol><hr /><h3 id="update-history">📜 Update History</h3><ul><li>v4.0: Major Logic Update (50% Draft Res), 5-Stack Detailer, Analog Finish.</li><li>v3.0: Introduced Person/Face Detailer separation.</li><li>v2.0: Added Hand Detailer &amp; Image Compare.</li><li>v1.0: Initial Q4/Q8 Switch Logic.</li></ul>Happy Creating! If this workflow saves you time, please leave a review and share your results! 🎨 <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/9e45e665-ba08-4480-bb6d-2c39a9310021/original=true/9e45e665-ba08-4480-bb6d-2c39a9310021.jpeg" />

Flux.1 Dev: GGUF The "Smart-Render" Pipeline (v4.0)

v3.0 UPDATE: Targeted Detailing &amp; Cinematic FinishThis update shifts focus from general enhancement to subject-specific precision.<ol><li>New Person/Face Detailer: Replaced the generic Detailer with a dedicated Person/Face detector. This ensures only your subject is refined, leaving the background consistent.</li><li>The Analog Finish: Added a Film Grain module (with toggle) to break the "AI-smoothness" and add realistic texture.</li><li>Workflow Optimization: Cleaned up group logic and wireless connections.</li></ol>

Files with SHA256: 284dedd1f6d878a1fdcac16b241b408c44861a621099d0440bd615334a372f2f