🎨 Qwen Image 2512 + SAM3 Face Inpainting Workflow&gt; Generate stunning portraits with automatic face detail enhancement — using single prompts or batch scheduling.<hr />## ✨ What This Workflow DoesThis ComfyUI workflow combines Qwen Image 2512 (a powerful text-to-image model) with SAM3 Smart Inpainter to:1. 🖼️ Generate high-quality images from text prompts2. 👤 Automatically find and enhance faces (man + woman) using AI segmentation3. 🔁 Support single prompt OR batch prompt scheduling for creative variations4. ⚡ Speed up generation with CacheDiT acceleration5. 🎯 Use advanced sampling for better resultsPerfect for: character portraits, fashion shoots, story illustration, face refinement, and creative batch generation.<hr />## ✨ What's New in V1.1### 🔁 Auto-Reset Counter (No More Manual Reset!)- Replaced the old <code>PrimitiveInt</code> node with my new AutoResetInt node- ✅ Automatically resets to <code>0</code> after each generation- ✅ Also resets if you cancel a generation- ✅ Want to run multiple batches? Just wait for the first image to finish — the counter resets itself!- 🎯 No more forgetting to reset — just click "Queue" and go### ⚙️ Model Format Switch- Added a simple toggle to choose your model format: - GGUF — for lower VRAM / quantized setups - FP8 / BF16 — for full quality on powerful GPUs- 🎯 Pick what works best for your hardware — no need to edit the workflow<hr />## 📦 Required Custom NodesInstall these in your <code>ComfyUI/custom_nodes/</code> folder:| Node Pack | Purpose | Link ||-----------|---------|------|| RES4LYF | Advanced sampler with better quality control | <a target="_blank" rel="ugc" href="https://github.com/ClownsharkBatwing/RES4LYF">https://github.com/ClownsharkBatwing/RES4LYF</a> || SAM3_SmartInpainter | Auto-detect &amp; inpaint faces using SAM3 | <a target="_blank" rel="ugc" href="https://github.com/majidfida/SAM3_SmartInpainter">https://github.com/majidfida/SAM3_SmartInpainter</a> || ComfyUI-CacheDiT | Speed up Qwen Image generation | <a target="_blank" rel="ugc" href="https://github.com/Jasonzzt/ComfyUI-CacheDiT">https://github.com/Jasonzzt/ComfyUI-CacheDiT</a> || comfyui_auto_prompt_schedule | Run multiple prompts in one batch | <a target="_blank" rel="ugc" href="https://github.com/ReinerBforartists/comfyui_auto_prompt_schedule">https://github.com/ReinerBforartists/comfyui_auto_prompt_schedule</a> |### Also Required:- ComfyUI-Easy-Sam3 or ComfyUI-RMBG (provides the <code>sam3</code> Python package for SAM3_SmartInpainter)- Qwen Image model files: <code>qwen_image_vae.safetensors</code>, <code>qwen_2.5_vl_7b_fp8_scaled.safetensors</code>, and your chosen checkpoint (e.g., <code>Qwen-Overcooked_fp8_V2.safetensors</code>)- SAM3 model: Download <code>sam3.pt</code> from <a target="_blank" rel="ugc" href="https://huggingface.co/Majidfida/Sam3/tree/main">https://huggingface.co/Majidfida/Sam3/tree/main</a> → place in <code>ComfyUI/models/sam3/</code><hr />## 🔧 How to Use### Step 1: Install &amp; Setup1. Clone all required node packs into <code>custom_nodes/</code>2. Install Python requirements (see each repo's README)3. Download models and place them in correct folders4. Restart ComfyUI### Step 2: Load the Workflow1. Drag the <code>.json</code> file into ComfyUI2. Wait for nodes to load (missing nodes will show red)### Step 3: Choose Your Mode#### 🎯 Single Prompt Mode (Default)- Set the toggle <code>true=Single Prompt | false=Batch Prompts</code> → true- Edit the prompt in the <code>⚡Single Prompt⚡</code> node- Click Queue Prompt#### 🔄 Batch Prompt Mode- Set the toggle → false- Edit prompts in the <code>Input Prompts for Schedule</code> node (format: <code>"0": "prompt text", "30": "next prompt", ...</code>)- Set <code>max_frames</code> in PromptSchedule node to control how many to generate- Click Queue Prompt — it will auto-run through your list### Step 4: Face Inpainting (Automatic)The workflow automatically:1. Finds the woman's face → enhances it with prompt: <code>"woman, exquisite detailed face..."</code>2. Finds the man's face → enhances it with prompt: <code>"man with dark beard, exquisite detailed face..."</code>3. Blends both back into the main image💡 Tip: To inpaint different subjects, chain more <code>SAM_SmartInpainter</code> nodes and change the <code>prompt</code> field (e.g., <code>"hat"</code>, <code>"eyes"</code>, <code>"scarf"</code>).### Step 5: Adjust Key Settings| Parameter | Recommended | What It Does ||-----------|-------------|--------------|| <code>min_mask_area</code> | <code>0.001</code> | Allows tiny faces to be detected || <code>crop_padding</code> | <code>32–128</code> | Extra space around face for better inpainting || <code>inpaint_resolution</code> | <code>1280</code> | Higher = more detail in repainted face || <code>denoise</code> | <code>0.3–0.5</code> | Lower = subtle changes, higher = more creative || <code>steps</code> | <code>6</code> | Qwen Image works well with low steps || <code>cfg</code> | <code>1.0</code> | Keep low for Qwen Image |---## 💡 Pro Tips🔹 Small or distant faces not detected?→ Lower <code>min_mask_area</code> to <code>0.0005</code> and <code>text_threshold</code> to <code>0.25</code>🔹 Inpainting looks blurry?→ Raise <code>inpaint_resolution</code> to <code>1280</code> or <code>1536</code>🔹 Edges look harsh?→ Use <code>blend_mode: feather</code> and increase <code>mask_blur</code>🔹 Want to inpaint more than faces?→ Change the SAM prompt to <code>"hat"</code>, <code>"logo"</code>, <code>"hands"</code> — it will find and repaint that area🔹 Batch generation too slow?→ Enable CacheDiT and use fp8 models for faster runs<hr />## ⚠️ Important Notes🔸 This workflow is designed for Qwen Image 2512 — other models may need parameter adjustments🔸 I do not monitor this post — I may not reply to comments or issues. This is shared for the community to use and improve.🔸 Always test with one image first before running large batches---## 🙏 Credits- Qwen Image 2512 — Alibaba Cloud / Qwen Team- SAM3 Model — Meta AI Research / Majidfida (HuggingFace)- RES4LYF — @ClownsharkBatwing- SAM3_SmartInpainter — @majidfida- CacheDiT — @Jasonzzt- Auto Prompt Schedule — @ReinerBforartists<hr />&gt; 🎨 Simple rule: Start with single prompt → get it perfect → then try batch mode for variations.&gt; If faces aren't found → lower <code>min_mask_area</code>. If results look soft → raise <code>inpaint_resolution</code>.Happy creating! ✨👫🖼️

Qwen Image 2512 + SAM3 Face Inpainting Workflow

## ✨ What's New in V1.1### 🔁 Auto-Reset Counter (No More Manual Reset!)- Replaced the old <code>PrimitiveInt</code> node with my new AutoResetInt node- ✅ Automatically resets to <code>0</code> after each generation- ✅ Also resets if you cancel a generation- ✅ Want to run multiple batches? Just wait for the first image to finish — the counter resets itself!- 🎯 No more forgetting to reset — just click "Queue" and go### ⚙️ Model Format Switch- Added a simple toggle to choose your model format: - GGUF — for lower VRAM / quantized setups - FP8 / BF16 — for full quality on powerful GPUs- 🎯 Pick what works best for your hardware — no need to edit the workflow---## 🧩 What This Workflow Does1. 🖼️ Generate high-quality images from text using Qwen Image 25122. 👤 Automatically find faces using SAM3 AI3. 🎨 Repaint only the face area for extra detail4. 🔁 Support single prompts OR batch prompt scheduling5. ⚡ Speed up with CacheDiT accelerationPerfect for: couple portraits, character art, face refinement, fashion shoots, and creative batch generation.---## 📦 What You Need to Install### Required Custom Nodes:| Node | Purpose | Link ||------|---------|------|| AutoResetInt | Auto-resetting counter (NEW!) | <a target="_blank" rel="ugc" href="https://github.com/majidfida/AutoResetInt">https://github.com/majidfida/AutoResetInt</a> || SAM3_SmartInpainter | Auto face detection + inpainting | <a target="_blank" rel="ugc" href="https://github.com/majidfida/SAM3_SmartInpainter">https://github.com/majidfida/SAM3_SmartInpainter</a> || RES4LYF | Advanced sampler for better quality | <a target="_blank" rel="ugc" href="https://github.com/ClownsharkBatwing/RES4LYF">https://github.com/ClownsharkBatwing/RES4LYF</a> || CacheDiT | Speed up Qwen Image generation | <a target="_blank" rel="ugc" href="https://github.com/Jasonzzt/ComfyUI-CacheDiT">https://github.com/Jasonzzt/ComfyUI-CacheDiT</a> || Auto Prompt Schedule | Run multiple prompts in one batch | <a target="_blank" rel="ugc" href="https://github.com/ReinerBforartists/comfyui_auto_prompt_schedule">https://github.com/ReinerBforartists/comfyui_auto_prompt_schedule</a> |### Also Required:- ComfyUI-Easy-Sam3 or ComfyUI-RMBG (provides the <code>sam3</code> Python package)- Qwen Image models: <code>qwen_image_vae.safetensors</code>, <code>qwen_2.5_vl_7b_fp8_scaled.safetensors</code>, and your chosen checkpoint- SAM3 model: Download <code>sam3.pt</code> from <a target="_blank" rel="ugc" href="https://huggingface.co/Majidfida/Sam3/tree/main">https://huggingface.co/Majidfida/Sam3/tree/main</a> → place in <code>ComfyUI/models/sam3/</code>---## 🔧 How to Use (Super Simple)### Step 1: Install &amp; Setup1. Clone all required nodes into <code>ComfyUI/custom_nodes/</code>2. Install Python requirements (see each repo's README)3. Download models and place them in the correct folders4. Restart ComfyUI### Step 2: Load the Workflow- Drag the <code>.json</code> file into ComfyUI- Wait for nodes to load (missing nodes will show red)### Step 3: Pick Your Settings| Setting | What to Do ||---------|------------|| Model Format Switch | Choose <code>GGUF</code> (low VRAM) or <code>FP8/BF16</code> (high quality) || Prompt Mode Toggle | <code>true</code> = single prompt, <code>false</code> = batch prompts || SAM Prompt | Type what to enhance: <code>"woman face"</code>, <code>"man beard"</code>, etc. || min_mask_area | Lower to <code>0.001</code> if faces are small or far away || inpaint_resolution | Keep at <code>1280</code> for sharp face detail |### Step 4: Generate!- Click Queue Prompt- Watch the magic happen ✨- For batch mode: edit prompts in the schedule node and set <code>max_frames</code>---## 💡 Pro Tips🔹 Small faces not detected? → Lower <code>min_mask_area</code> to <code>0.0005</code> and <code>text_threshold</code> to <code>0.25</code>🔹 Inpainted face looks blurry? → Raise <code>inpaint_resolution</code> to <code>1280</code> or <code>1536</code>🔹 Edges look harsh? → Use <code>blend_mode: feather</code> and increase <code>mask_blur</code>🔹 Want to inpaint more than faces? → Change the SAM prompt to <code>"hat"</code>, <code>"eyes"</code>, <code>"hands"</code> — it will find and repaint that area🔹 Running multiple batches? → Just wait for the first image to finish — AutoResetInt resets itself automatically!---## ⚠️ Important Notes🔸 SAM3 only — This workflow has only been tested with <code>sam3.pt</code> 🔸 I do not monitor this post — I may not reply to comments or issues. This is shared for the community to use and improve. 🔸 Always test with one image first before running large batches 🔸 GGUF vs FP8: GGUF uses less VRAM but may have slightly lower quality. FP8/BF16 gives best results if your GPU can handle it.---## 🙏 Credits- Qwen Image 2512 — Alibaba Cloud / Qwen Team - SAM3 Model — Meta AI Research / Majidfida (HuggingFace) - AutoResetInt — @majidfida - SAM3_SmartInpainter — @majidfida - RES4LYF — @ClownsharkBatwing - CacheDiT — @Jasonzzt - Auto Prompt Schedule — @ReinerBforartists ---&gt; 🎯 Simple rule: Start with one prompt → get settings perfect → then try batch mode. &gt; If faces aren't found → lower <code>min_mask_area</code>. If results look soft → raise <code>inpaint_resolution</code>.Happy creating! 🖼️✨👫

Files with SHA256: 9c3be9b11eb7d6c4c88c058aa0144c5ba029b75b5364ef0e55aeda72dc059c7b