Davemane42's Regional Conditioning Nodes for ComfyUI

Updated: November 22, 2025 Version: 3.1 - Enhanced with ALL-IN-ONE nodes featuring inline prompts!

Visual interface for regional conditioning in ComfyUI. Draw boxes on a canvas and assign different prompts to each region - perfect for precise control over image composition.

🌟 NEW! Enhanced Easy-Mode Nodes (RECOMMENDED)

🎨 Regional Prompter (SD/SDXL - Easy!) and 🎨 Regional Prompter (Flux/Chroma - Easy!)

✅ Type prompts directly in the node - No external CLIP Text Encode nodes needed! ✅ Simple workflow: Checkpoint → CLIP → Regional Prompter → Sampler ✅ Same visual box drawing you love ✅ Flux-optimized with perfect mask strength (0.8) and feathering ✅ Up to 4 regions + background

Perfect for beginners and quick workflows!

---

✨ Features

• 🎨 Visual Box Drawing Interface - Color-coded regions with interactive canvas
• 📐 Grid Overlay - 64px alignment grid for precise positioning
• 💪 Strength Control - Per-region strength adjustment (0.0-10.0)
• 🔄 Right-Click Menu - Easy add/remove/swap conditioning layers
• ✅ Automatic Boundary Clipping - Smart handling of out-of-bounds regions
• 🛡️ Comprehensive Error Handling - Helpful validation and fallback behavior
• 🚀 Modern Model Support - Area-based AND mask-based conditioning

---

📦 Installation

Method 1: ComfyUI Manager (Recommended)

1. Open ComfyUI Manager
2. Search for "Regional Conditioning"
3. Click Install

Method 2: Manual Installation

cd ComfyUI/custom_nodes/
git clone https://github.com/Davemane42/ComfyUI_RegionalConditioning
# Restart ComfyUI

Note: JavaScript files are automatically loaded from the js/ folder - no manual copying needed!

---

🎯 Model Compatibility

✅ MultiAreaConditioning (Area-based)

Compatible Models:

• ✅ Stable Diffusion 1.5
• ✅ Stable Diffusion 2.x
• ✅ SDXL (Stable Diffusion XL)

Technology: Uses ComfyUI's ConditioningSetArea internally Best For: Traditional Stable Diffusion workflows with rectangular regions

✅ MultiAreaConditioningMask (Mask-based)

Compatible Models:

• ✅ Flux (all variants)
• ✅ Chroma (Chroma1-Radiance, etc.)
• ✅ SD3 / SD3.5 (Stable Diffusion 3)
• ✅ Any model supporting ConditioningSetMask

Technology: Automatically converts boxes to binary masks Best For: Modern models that require mask-based conditioning

❓ Experimental Support

• 🟡 Qwen-Image (generation model) - May work with standard CONDITIONING
• ❌ WAN 2.2 - Not currently supported (video-focused architecture)

---

🚀 Quick Start Example

🌟 NEW! Easy Mode (Recommended)

Using Regional Prompter (inline prompts):

1. Load your checkpoint
2. Add 🎨 Regional Prompter (SD/SDXL) OR 🎨 Regional Prompter (Flux/Chroma)
3. Connect CLIP from checkpoint to the Regional Prompter
4. Type in Background Prompt: "photo of a mountain landscape"
5. Draw a box on canvas, type in Region 1 Prompt: "majestic tiger"
6. Draw another box, type in Region 2 Prompt: "berry bush"
7. Connect conditioning output to your sampler
8. Done! 🎉

That's it! No external CLIP Text Encode nodes, no complex wiring!

Advanced Mode (Original Nodes)

Using MultiAreaConditioning (requires external CLIP Text Encode):

1. Add MultiAreaConditioning (for SDXL) OR MultiAreaConditioningMask (for Flux)
2. Connect conditioning0 → CLIPTextEncode("photo of a mountain landscape")
3. Draw a box, connect conditioning1 → CLIPTextEncode("majestic tiger")
4. Draw another box, connect conditioning2 → CLIPTextEncode("berry bush")
5. Connect output to your sampler

Both methods give the same result: A mountain landscape with a tiger and berry bush positioned exactly where you drew the boxes!

---

📚 Node Reference

🌟 Regional Prompter (SD/SDXL - Easy!) ← RECOMMENDED

The easiest way to use regional conditioning with SDXL!

Inputs:

• clip - CLIP model from your checkpoint
• background_prompt - Text box for overall scene (multiline)
• region1_prompt - Text box for first region (multiline)
• region2_prompt - Text box for second region (optional)
• region3_prompt - Text box for third region (optional)
• region4_prompt - Text box for fourth region (optional)

Outputs:

• conditioning - Ready to connect to sampler!
• width, height - Canvas dimensions

How it works: Just type your prompts and draw boxes - the node handles CLIP encoding internally!

Perfect for: Quick workflows, beginners, anyone who wants simplicity

---

🌟 Regional Prompter (Flux/Chroma - Easy!) ← RECOMMENDED FOR FLUX

The easiest way to use regional conditioning with Flux!

Inputs:

• clip - CLIP model from your checkpoint
• width, height - Output dimensions
• background_prompt - Text box for overall scene (multiline)
• region1_prompt - Text box for first region (multiline)
• region2_prompt - Text box for second region (optional)
• region3_prompt - Text box for third region (optional)
• region4_prompt - Text box for fourth region (optional, limit 3-4 for best results)
• flux_optimize - Enable Flux optimizations (default: ON)

Output:

• conditioning - Ready to connect to sampler!

Flux Optimizations:

• ✅ Softened mask strength (0.8 vs 1.0) for better blending
• ✅ Gentle feathering at region edges
• ✅ Automatic region limiting (warns if >4 regions)

Tips:

• Use 3-4 regions maximum for reliable results
• Enable Flux Optimize for best quality
• Increase CFG to 5-7 (vs typical 3-5)
• Draw larger regions for better control

Perfect for: Flux workflows, Chroma, SD3, modern models

---

📚 Advanced Node Reference

MultiAreaConditioning (Area-based for SD/SDXL)

Inputs:

• conditioning0 - Base/background conditioning (fullscreen if no box drawn)
• conditioning1+ - Regional conditioning (add more via right-click menu)

Outputs:

• conditioning - Combined regional conditioning
• resolutionX - Canvas width (INT)
• resolutionY - Canvas height (INT)

Widgets:

• resolutionX/Y - Canvas dimensions (auto-synced from latent)
• index - Select which region to edit
• x, y - Region position
• width, height - Region dimensions
• strength - Region influence (0.0-10.0)

Right-Click Menu:

• Insert input above/below current
• Swap with input above/below
• Remove currently selected input
• Remove all unconnected inputs

---

MultiAreaConditioningMask (Mask-based for Flux/Chroma)

Inputs:

• conditioning0 - Base/background conditioning
• conditioning1+ - Regional conditioning (add more via right-click menu)
• width - Output width in pixels (default: 1024)
• height - Output height in pixels (default: 1024)

Output:

• conditioning - Combined mask-based conditioning

Widgets: (same as area-based)

• index - Select which region to edit
• x, y - Region position
• width, height - Region dimensions
• strength - Region influence (0.0-10.0)

How It Works: Boxes are automatically converted to binary masks behind the scenes. You draw visually, the node handles the mask generation!

---

ConditioningUpscale

Upscale conditioning areas by a scalar factor - perfect for hi-res fix workflows.

Inputs:

• conditioning - Conditioning to upscale
• scalar - Scaling factor (1-100, default: 2)

Output:

• conditioning - Upscaled conditioning

Example: 512x512 regions × scalar=2 → 1024x1024 regions

---

ConditioningStretch

Stretch/resize conditioning areas to fit new dimensions - more flexible than upscale.

Inputs:

• conditioning - Conditioning to resize
• resolutionX/Y - Current/original dimensions
• newWidth/Height - Target dimensions

Output:

• conditioning - Resized conditioning

Example: Transform regions from 512x512 to 1024x768 (proportional scaling)

---

MultiLatentComposite

Visual interface for compositing multiple latents with positioning and feathering.

Inputs:

• samples_to - Base/destination latent canvas
• samples_from0+ - Latents to composite (add more via right-click menu)

Output:

• LATENT - Composited latent

Widgets:

• index - Select which latent to edit
• x, y - Position in pixels
• feather - Blend distance (0 = hard edge, higher = smooth blend)

Features:

• ✅ Automatic bounds checking and clipping
• ✅ Channel compatibility validation
• ✅ Smart feathering with gradient masks
• ✅ Detailed info/warning messages

Fixed: The "no bounds checking" issue mentioned in original README is now resolved!

---

🔧 Advanced Tips

Fullscreen/Background Conditioning

Set a region to x=0, y=0, width=canvas_width, height=canvas_height to make it fullscreen. This is perfect for background prompts.

Grid Alignment

The 64px grid overlay helps you align regions to latent space boundaries for optimal results.

Strength Values

• 1.0 - Normal strength (default)
• 0.5 - Subtle influence
• 2.0+ - Strong influence
• 0.0 - Effectively disabled

Multiple Regions

• Use right-click menu to add as many conditioning inputs as needed
• Each region can have different prompts, positions, and strengths
• Regions are processed in order and can overlap

---

🐛 Error Handling

This modernized version includes extensive error handling:

• ✅ Validation - All inputs validated with helpful error messages
• ✅ Fallback Behavior - Graceful degradation if metadata missing
• ✅ Bounds Checking - Automatic clipping for out-of-bounds regions
• ✅ Type Checking - Validates tensor shapes and data types
• ✅ Helpful Messages - Clear ⚠️ warnings and ℹ️ info in console

Common Issues:

• "Workflow metadata missing" → Save and reload your workflow
• "Region has zero width/height" → Check your box dimensions
• "Beyond canvas width/height" → Region position is out of bounds (auto-clipped)

---

📝 Changelog

Version 3.0 (November 2025)

• ✨ NEW: MultiAreaConditioningMask for Flux/Chroma support
• ✨ NEW: Comprehensive error handling and validation
• ✨ NEW: Bounds checking for MultiLatentComposite (fixes known issue)
• ✨ NEW: Tooltips and descriptions on all inputs
• 🔧 FIX: Modernized JavaScript loading (no more file copying!)
• 🔧 FIX: Removed debug console.log statements
• 🔧 FIX: Fixed graph reference bug in background drawing
• 📚 DOCS: Complete compatibility matrix and examples
• 🚀 PERF: Improved tensor operations and validation

Version 2.4 (Original)

• Visual area conditioning interface
• MultiLatentComposite with feathering
• ConditioningUpscale and ConditioningStretch utilities

---

🤝 Contributing

Found a bug? Have a feature request? Open an issue

---

📄 License

GLWT (Good Luck With That) Public License See LICENSE file for details.

---

🙏 Credits

Original Author: Davemane42#0042 Modernization: 2025-11-22 update for ComfyUI 0.3.71+

Special Thanks:

• ComfyUI community for testing and feedback
• Contributors who reported issues and suggested improvements

---

📸 Examples

MultiAreaConditioning Workflow

Traditional area-based regional conditioning for SDXL

MultiLatentComposite Workflow

Visual latent compositing with feathering

---

Enjoy precise regional control in your ComfyUI workflows! 🎨