<hr /><h1 id="flux-2-godmod-by-velopilot">⚡ FLUX 2 GODMOD — by VeloPilot</h1>👋 Created by VeloPilot — one-graph FLUX.2 Klein editor: text-to-image · editing · inpaint · outpaint · multi-reference<ul><li>🌐 <a target="_blank" rel="ugc" href="https://civarchive.com/user/Velopilot">CivitAI Profile</a></li><li>🐙 <a target="_blank" rel="ugc" href="https://github.com/comfyanonymous/ComfyUI">GitHub: ComfyUI</a></li><li>🤗 <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/flux2-klein-9B">FLUX.2 Klein 9B (Comfy-Org)</a></li><li>📖 <a target="_blank" rel="ugc" href="https://docs.comfy.org/tutorials/flux/flux-2-klein">Official ComfyUI FLUX.2 Klein guide</a></li></ul><hr /><h2 id="overview">📋 Overview</h2>One workflow that covers every FLUX.2 Klein task. You never rewire anything — you flip toggles.<ol><li>Step 1 — Load your photo into <code>LOAD IMAGE 1</code> (plus optional reference images 2 and 3)</li><li>Step 2 — Prompt — pick a ready task preset (character swap / clothing / pose) or write your own, optionally rewritten by a local LLM</li><li>Step 3 — Choose a mode — T2I, edit, inpaint or outpaint, all on one node</li><li>Step 4 — Run — masks can be painted by hand, generated by SAM3 text prompt, or both at once</li></ol>Mode Toggles on the <code>Sampler</code> node What it does 🖼️ Text → Image all toggles OFF clean generation at the selected aspect ratio ✏️ Edit <code>ENABLE EDIT MODE</code> character swap, clothing, pose, object removal 🖌️ Inpaint <code>ENABLE EDIT MODE</code> + <code>ENABLE INPAINT MODE</code> + mask repaint only the masked area, denoise slider active 🔲 Outpaint <code>ENABLE EDIT MODE</code> + <code>ENABLE OUTPAINTING MODE</code> extend the canvas by N pixels on any side<blockquote>⚠️ FIRST: install all custom nodes via ComfyUI Manager, then load the workflow.</blockquote><hr /><h2 id="features">✨ Features</h2><ul><li>🎚️ Everything on one node — modes, steps, denoise, expand values live on the <code>Sampler</code> subgraph, no digging around the canvas</li><li>🧩 Three image slots — main photo + two reference images (style, clothing, object) fed through <code>ReferenceLatent</code></li><li>🎯 SAM3 masking by text — type <code>jacket</code>, <code>hair</code>, <code>background</code> and get a mask, no brush needed</li><li>🖌️ Hand mask + SAM3 combined — the two masks are added together</li><li>🧠 Optional LLM prompt enhancer — rewrites a short request into a full FLUX.2 prompt, can read the input image</li><li>📝 Task presets — character swap / clothing change / pose change templates that keep identity and framing locked</li><li>🧰 Preprocessors on board — DWPose, Depth Anything V2, Canny, HED, M-LSD, LineArt, auto face crop</li><li>🔀 One-click bypass panels — disable image slots or the LLM group without touching links</li></ul><hr /><h2 id="required-custom-nodes">🔧 Required Custom Nodes</h2>Node pack GitHub Used for rgthree-comfy <a target="_blank" rel="ugc" href="https://github.com/rgthree/rgthree-comfy">rgthree/rgthree-comfy</a> Power Lora Loader, Fast Groups Bypasser, Any Switch, Image Comparer, Image Resize ComfyUI-KJNodes <a target="_blank" rel="ugc" href="https://github.com/kijai/ComfyUI-KJNodes">kijai/ComfyUI-KJNodes</a> JoinStrings, GrowMaskWithBlur, Sage Attention patch ComfyUI-Easy-Use <a target="_blank" rel="ugc" href="https://github.com/yolain/ComfyUI-Easy-Use">yolain/ComfyUI-Easy-Use</a> easy boolean / float / imageSize / isMaskEmpty, cache &amp; VRAM cleanup ComfyUI-Easy-Sam3 <a target="_blank" rel="ugc" href="https://github.com/yolain/ComfyUI-Easy-Sam3">yolain/ComfyUI-Easy-Sam3</a> SAM3 model loader, SAM3 image segmentation (text-prompt masks) ComfyUI-Crystools <a target="_blank" rel="ugc" href="https://github.com/crystian/ComfyUI-Crystools">crystian/ComfyUI-Crystools</a> Switch any / Switch conditioning (all mode toggles) comfyui_controlnet_aux <a target="_blank" rel="ugc" href="https://github.com/Fannovel16/comfyui_controlnet_aux">Fannovel16/comfyui_controlnet_aux</a> DWPose, Depth Anything V2, Canny, HED, M-LSD, LineArt ComfyUI_Comfyroll_CustomNodes <a target="_blank" rel="ugc" href="https://github.com/Suzie1/ComfyUI_Comfyroll_CustomNodes">Suzie1/ComfyUI_Comfyroll_CustomNodes</a> CR Image Border (outpaint canvas) ComfyUI-AutoCropFaces <a target="_blank" rel="ugc" href="https://github.com/liusida/ComfyUI-AutoCropFaces">liusida/ComfyUI-AutoCropFaces</a> automatic face crop for reference images was-node-suite-comfyui <a target="_blank" rel="ugc" href="https://github.com/WASasquatch/was-node-suite-comfyui">WASasquatch/was-node-suite-comfyui</a> Image Blank ComfyLiterals <a target="_blank" rel="ugc" href="https://github.com/M1kep/ComfyLiterals">M1kep/ComfyLiterals</a> Int<blockquote>✅ <code>TextGenerate</code>, <code>RegexReplace</code>, <code>ReferenceLatent</code>, <code>ResolutionSelector</code>, <code>SetLatentNoiseMask</code>, <code>PrimitiveStringMultiline</code> and <code>MarkdownNote</code> are built into ComfyUI core — just keep ComfyUI up to date.</blockquote><hr /><h2 id="required-models-and-files">📦 Required Models &amp; Files</h2><h3 id="diffusion_models">🧠 diffusion_models</h3><ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/black-forest-labs/FLUX.2-klein-9b-fp8">flux-2-klein-9b-fp8.safetensors</a> — distilled, the one this workflow is tuned for → <code>models/diffusion_models/</code></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/black-forest-labs/FLUX.2-klein-base-9b-fp8">flux-2-klein-base-9b-fp8.safetensors</a> — optional base version → <code>models/diffusion_models/</code></li></ul><blockquote>🔒 The BFL repos are gated — accept the license on HuggingFace before downloading.</blockquote><h3 id="text_encoders">🔤 text_encoders</h3><ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/flux2-klein-9B/resolve/main/split_files/text_encoders/qwen_3_8b_fp8mixed.safetensors">qwen_3_8b_fp8mixed.safetensors</a> — main text encoder → <code>models/text_encoders/</code></li><li>(optional) any uncensored Qwen3 / Gemma text encoder for the LLM prompt enhancer group → <code>models/text_encoders/</code></li></ul><blockquote>💡 The screenshots use finetuned/abliterated text encoders. They are not required — point both loaders at the stock <code>qwen_3_8b</code> and everything works.</blockquote><h3 id="vae">🌀 vae</h3><ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/flux2-dev/resolve/main/split_files/vae/flux2-vae.safetensors">flux2-vae.safetensors</a> → <code>models/vae/</code></li></ul><h3 id="sam3-(text-prompt-masking)">🎯 sam3 (text-prompt masking)</h3><ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/yolain/sam3-safetensors/blob/main/sam3-fp16.safetensors">sam3-fp16.safetensors</a> or the original <a target="_blank" rel="ugc" href="https://huggingface.co/facebook/sam3">facebook/sam3</a> → <code>models/sam3/</code></li></ul><h3 id="preprocessors-(auto-download)">🧍 Preprocessors (auto-download)</h3><ul><li><code>yolox_l.onnx</code> and <code>dw-ll_ucoco_384_bs5.torchscript.pt</code> — pulled automatically by comfyui_controlnet_aux on first DWPose run</li><li>Depth Anything V2 / HED / LineArt / M-LSD checkpoints — same, downloaded on demand</li></ul><h3 id="loras">🎭 loras</h3><ul><li>Any FLUX.2 Klein LoRA of your choice — drop it into the Power Lora Loader, and put its trigger words into the <code>Lora keywords</code> node. Nothing is required to run the workflow.</li></ul><hr /><h2 id="model-folder-structure">📁 Model Folder Structure</h2><pre><code>📂 ComfyUI/
├── 📂 custom_nodes/
│ ├── rgthree-comfy/
│ ├── ComfyUI-KJNodes/
│ ├── ComfyUI-Easy-Use/
│ ├── ComfyUI-Easy-Sam3/
│ ├── ComfyUI-Crystools/
│ ├── comfyui_controlnet_aux/
│ ├── ComfyUI_Comfyroll_CustomNodes/
│ ├── ComfyUI-AutoCropFaces/
│ ├── was-node-suite-comfyui/
│ └── ComfyLiterals/
└── 📂 models/
 ├── 📂 diffusion_models/
 │ └── flux-2-klein-9b-fp8.safetensors
 ├── 📂 text_encoders/
 │ ├── qwen_3_8b_fp8mixed.safetensors
 │ └── (optional) your-uncensored-LLM.safetensors
 ├── 📂 vae/
 │ └── flux2-vae.safetensors
 ├── 📂 sam3/
 │ └── sam3-fp16.safetensors
 └── 📂 loras/
 └── your_flux2_klein_lora.safetensors
</code></pre><hr /><h2 id="how-to-use">🎬 How to use</h2><h3 id="1-text-image">1️⃣ Text → Image</h3>Leave every toggle off, pick an aspect ratio in <code>ResolutionSelector</code>, write a prompt, run. Denoise is forced to <code>1.0</code> automatically.<h3 id="2-edit-character-swap-clothing-pose">2️⃣ Edit / character swap / clothing / pose</h3>Load a photo into <code>LOAD IMAGE 1</code>, enable <code>ENABLE EDIT MODE</code>, then take a preset from the PROMTPS FOR HELP panel:<ul><li>CHARACTER SWAP — new person, same pose, framing, light and background</li><li>CLOTHING — new outfit, same person and scene</li><li>POSE — same person and outfit, new body position</li></ul>Paste the preset into <code>User prompt</code> and finish the sentence after <code>Specific request:</code> with what you actually want.<h3 id="3-inpaint">3️⃣ Inpaint</h3>Enable <code>ENABLE EDIT MODE</code> + <code>ENABLE INPAINT MODE</code>, then give it a mask in one of three ways:<ul><li>paint it on the mask loader node and connect <code>MASK</code> → <code>mask</code> of <code>IMAGE 1 PROCESS</code></li><li>type an object name into the SAM3 prompt inside <code>IMAGE 1 PROCESS</code> (<code>jacket</code>, <code>hair</code>, <code>sign</code>…)</li><li>both at once — the masks are summed</li></ul><code>Denoise</code> is the only mode where the slider matters: <code>0.5–0.7</code> for a soft retouch, <code>0.9–1.0</code> for a full repaint.<h3 id="4-outpaint">4️⃣ Outpaint</h3>Enable <code>ENABLE EDIT MODE</code> + <code>ENABLE OUTPAINTING MODE</code> and set <code>Expand top / bottom / left / right</code> in pixels. White borders are added, and the model paints them in with matching perspective, light and grain.<hr /><h2 id="recommended-settings">⚙️ Recommended settings</h2>Parameter Value Note Steps <code>4</code> Klein 9B is distilled — more steps give a waxy, over-cooked look CFG <code>1.0</code> distilled model Guidance <code>4.0</code> drop to <code>2.5–3.5</code> if edits drift away from the original Sampler / Scheduler <code>euler</code> / <code>simple</code> avoid ancestral samplers Denoise <code>1.0</code> everywhere except inpaint forced automatically Aspect ratio match the source photo the input is center-cropped to the selected ratio<hr /><h2 id="tips">💡 Tips</h2><ul><li>Reference images 2 and 3 are for style, an outfit or an object you want carried into the main photo. Turn their groups off in the bypass panel when unused — they cost VRAM.</li><li>LLM enhancer is worth it for short requests. A long, precise prompt is still better written by hand — bypass the group and your text goes to the model untouched.</li><li>Mask stuck on the image? ComfyUI saves painted masks into <code>input/clipspace/</code> and re-points the loader at that file. Pick the original file in the node dropdown to reset it — that is why masking lives on a separate loader here, so the source photo never gets overwritten.</li><li>Preprocessors (pose / depth / canny / lineart) are optional guides — enable one at a time, they are heavy.</li><li>Low VRAM: bypass the LLM group and reference slots 2–3, and keep the resolution at 1MP.</li></ul><hr /><h2 id="report-issues">🐛 Report Issues</h2>Before reporting, please:<ol><li>Update ComfyUI to the latest version — <a target="_blank" rel="ugc" href="https://docs.comfy.org/installation/update_comfyui">Update Guide</a></li><li>Update all custom nodes via ComfyUI Manager</li><li>Check that every model sits in the folder listed above</li><li>Tell me your GPU, VRAM and the exact error text from the console</li></ol><hr />Made by VeloPilot — AI Content Creator This workflow is provided as-is for educational and creative purposes.The author assumes no responsibility for any content generated with it.Users are solely responsible for ensuring their use of all includedmodels, LoRAs, and custom nodes complies with their respective licensesand terms of distribution. Third-party models retain their originallicenses - refer to each model's page on Hugging Face or GitHub forfull terms.

FLUX 2 GODMOD — by VeloPilot

One workflow that covers every FLUX.2 Klein task. You never rewire anything — you flip toggles.<ol><li>Step 1 — Load your photo into <code>LOAD IMAGE 1</code> (plus optional reference images 2 and 3)</li><li>Step 2 — Prompt — pick a ready task preset (character swap / clothing / pose) or write your own, optionally rewritten by a local LLM</li><li>Step 3 — Choose a mode — T2I, edit, inpaint or outpaint, all on one node</li><li>Step 4 — Run — masks can be painted by hand, generated by SAM3 text prompt, or both at once</li></ol>

Files with SHA256: 3f40e0328e9db2c645ffcef8f66ad2814f19fa058fefba66c3eb3abb977a6502