arrangebpy

Automatic layout of nodes for Blender node trees - A clean, library-focused Python module implementing multiple graph layout algorithms.

Ported from the excellent node-arrange Blender add-on, redesigned as a standalone module for use in other add-ons and tools.

Features

🎨 Multiple Layout Algorithms

• Sugiyama: Hierarchical layout with minimal edge crossings (default)
• Topological: Fast layered layout for quick results
• Grid: Regular grid arrangement, great for organization
• Orthogonal: Clean orthogonal edge routing for presentations

✨ Complete Sugiyama Implementation

• Hierarchical node layout with minimal edge crossings
• Frame-aware layout (respects Blender's node frames)
• Multi-input socket handling
• Reroute node optimization

🎛️ Highly Configurable

• 5 layout directions: Balanced, Left-Up, Right-Down, etc.
• 3 socket alignment modes: None, Moderate, Full
• Node stacking: Automatically stack collapsed math nodes
• Iterative refinement: Better layouts for complex frame hierarchies
• Adjustable spacing: Control horizontal and vertical spacing

🔧 Clean API Design

• Unified layout() function for all algorithms
• Settings-based configuration (no global state)
• Type-safe with full type hints
• Well-documented with examples
• Designed for embedding in other add-ons

Installation

pip install arrangebpy

Or with uv:

uv add arrangebpy

Quick Start

from arrangebpy import layout, LayoutSettings

# Simple usage with defaults (uses Sugiyama)
layout(node_tree)

# Choose a specific algorithm
layout(node_tree, algorithm="topological")  # Fast layout
layout(node_tree, algorithm="grid")         # Grid layout
layout(node_tree, algorithm="orthogonal")   # Orthogonal edges

# Custom settings
from arrangebpy import LayoutSettings
settings = LayoutSettings(
    direction="BALANCED",
    socket_alignment="MODERATE",
    horizontal_spacing=60.0,
    vertical_spacing=30.0,
)
layout(node_tree, algorithm="sugiyama", settings=settings)

Legacy API (Still Supported)

from arrangebpy import sugiyama_layout, LayoutSettings

# Direct function calls still work
sugiyama_layout(node_tree)
sugiyama_layout(node_tree, LayoutSettings(horizontal_spacing=60.0))

Layout Algorithms

Sugiyama (Hierarchical)

Best for most node trees. Creates a hierarchical left-to-right layout with minimal edge crossings.

from arrangebpy import layout, LayoutSettings

layout(ntree, algorithm="sugiyama", settings=LayoutSettings(
    direction="BALANCED",
    socket_alignment="FULL",
    stack_collapsed=True,  # Stack collapsed math nodes
    align_top_layer=True   # Align input/output at top
))

Special Feature - Flat Top Layout:

# Align source and sink nodes (input/output) at Y=0, push others below
layout(ntree, algorithm="sugiyama", settings=LayoutSettings(
    align_top_layer=True  # Clean flat top with I/O aligned
))

Use when: You want the highest quality layout with minimal crossings (shader trees, geometry nodes, etc.)

Topological (Fast Layered)

Quick and simple layered layout without crossing reduction. Much faster than Sugiyama.

from arrangebpy import layout, TopologicalSettings

layout(ntree, algorithm="topological", settings=TopologicalSettings(
    horizontal_spacing=60.0,
    sort_by_degree=True  # Sort nodes by connection count
))

# For perfectly flat horizontal layouts (all nodes at Y=0)
layout(ntree, algorithm="topological", settings=TopologicalSettings(
    flatten=True  # Perfect for simple linear chains
))

Use when: You need quick layouts during development, have very large graphs, or want a perfectly flat horizontal layout.

Grid

Arranges nodes in a regular grid pattern, optionally grouping by type or cluster.

from arrangebpy import layout, GridSettings

layout(ntree, algorithm="grid", settings=GridSettings(
    columns=5,
    grouping="TYPE",  # Group by node type
    cell_width=250.0
))

Use when: You want to organize collections of nodes or create inventory-style layouts.

Orthogonal (Clean Edges)

Uses Sugiyama for node placement but routes edges with only horizontal/vertical segments.

from arrangebpy import layout, OrthogonalSettings

layout(ntree, algorithm="orthogonal", settings=OrthogonalSettings(
    horizontal_spacing=80.0,  # More space for routing
    socket_alignment="FULL"
))

Use when: You need professional-looking layouts for presentations or documentation.

Usage Examples

Default Layout (Sugiyama)

from arrangebpy import layout

# Uses sensible defaults
layout(material.node_tree)

Shader Node Trees (with stacking)

from arrangebpy import LayoutSettings
from arrangebpy.arrange.sugiyama import sugiyama_layout

settings = LayoutSettings(
    # Stack collapsed math nodes
    stack_collapsed=True,
    stack_margin_y_factor=0.4,
    
    # Full socket alignment for cleaner connections
    socket_alignment="FULL",
    
    # Balanced direction
    direction="BALANCED",
    
    # Tight spacing
    horizontal_spacing=50.0,
    vertical_spacing=20.0,
)

sugiyama_layout(shader_node_tree, settings)

Geometry Node Trees

settings = LayoutSettings(
    # More spacing for larger nodes
    horizontal_spacing=70.0,
    vertical_spacing=30.0,
    
    # Left-to-right flow
    direction="RIGHT_DOWN",
    
    # Moderate socket alignment
    socket_alignment="MODERATE",
    
    # High quality layout
    iterations=20,
    crossing_reduction_sweeps=48,
)

sugiyama_layout(geometry_node_tree, settings)

Configuration

All configuration through the LayoutSettings dataclass:

from arrangebpy import LayoutSettings

settings = LayoutSettings(
    # Spacing
    horizontal_spacing=50.0,      # Horizontal spacing between columns
    vertical_spacing=25.0,        # Vertical spacing between nodes
    
    # Layout algorithm
    direction="BALANCED",         # Layout direction
    socket_alignment="MODERATE",  # Socket alignment mode
    iterations=20,                # BK algorithm refinement iterations
    crossing_reduction_sweeps=24, # Edge crossing minimization passes
    
    # Features
    add_reroutes=True,           # Add reroute nodes for clean routing
    stack_collapsed=False,       # Stack collapsed math nodes
    stack_margin_y_factor=0.5,   # Spacing factor for stacks (0-1)
)

Direction Options

• "BALANCED" - Combines all 4 directions (default, best results)
• "LEFT_DOWN" - Top-left alignment
• "RIGHT_DOWN" - Bottom-right alignment
• "LEFT_UP" - Bottom-left alignment
• "RIGHT_UP" - Top-right alignment

Socket Alignment Options

• "NONE" - Only align node tops (fastest)
• "MODERATE" - Smart alignment based on node heights (default)
• "FULL" - Always align sockets (cleanest connections)

Use in Blender Add-ons

import bpy
from arrangebpy import LayoutSettings
from arrangebpy.arrange.sugiyama import sugiyama_layout

class MY_OT_ArrangeNodes(bpy.types.Operator):
    bl_idname = "node.my_arrange"
    bl_label = "Arrange Nodes"
    
    def execute(self, context):
        ntree = context.space_data.edit_tree
        
        # Configure based on node tree type
        if ntree.bl_idname == 'ShaderNodeTree':
            settings = LayoutSettings(
                stack_collapsed=True,
                socket_alignment="FULL",
            )
        else:
            settings = LayoutSettings()
        
        sugiyama_layout(ntree, settings)
        return {'FINISHED'}

How It Works

Implements the Sugiyama hierarchical graph layout framework:

1. Graph Construction - Build directed graph from node tree
2. Ranking - Assign nodes to horizontal layers
3. Crossing Minimization - Reduce edge crossings via median heuristic
4. Coordinate Assignment - Place nodes using Brandes-Köpf algorithm
5. Edge Routing - Add bend points for clean connections
6. Realization - Apply positions back to Blender

Architecture

arrangebpy/
├── settings.py              # Configuration
├── arrange/
│   ├── sugiyama.py         # Main orchestration
│   ├── graph.py            # Graph construction
│   ├── ranking.py          # Rank assignment
│   ├── ordering.py         # Crossing minimization
│   ├── placement/
│   │   ├── bk.py           # Brandes-Köpf algorithm
│   │   └── linear_segments.py
│   ├── coordinates.py      # X-coordinate assignment  
│   ├── routing.py          # Edge routing
│   ├── stacking.py         # Node stacking
│   └── ...
└── utils.py

Development

git clone https://github.com/BradyAJohnston/arrangebpy.git
cd arrangebpy

# Install
uv sync

# Run tests
uv run pytest tests/

License

GPL-3.0-or-later

Credits

Based on node-arrange by Leonardo Pike-Excell

Implements algorithms from Sugiyama et al., Brandes & Köpf, and others.