nodes/aTanguay-ComfyUI_Detonate
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
26 Commits 3 Branches 0 Tags
4cb409e2cbcd9da8982ecf7748850830fa595be7
Commit Graph
2 Commits
Author SHA1 Message Date
Claude e15b4be146 feat: Tier 5 - Interactive Masking (RotoBezier, v0.7.0) 
MAJOR FEATURE: Professional rotoscoping tools with interactive Bezier spline drawing!

New Nodes (2):
- RotoBezier: Interactive Bezier spline drawing for mask creation
- RotoBezier From Image: Auto-dimension rotoscoping (matches input image)

Key Features:
- Web-based interactive drawing widget (HTML5 Canvas)
- de Casteljau's algorithm for numerically stable Bezier evaluation
- Supersampling anti-aliasing (1-16×) for film-quality smooth edges
- Distance field feathering with smoothstep falloff (superior to blur)
- Multiple splines support (additive)
- Keyboard shortcuts: Enter/Delete/Escape
- JSON-based widget-to-backend data exchange

Technical Implementation:
- Frontend: web/js/detonate_roto_bezier.js (interactive widget)
  - BezierPoint, BezierSpline, BezierDrawingWidget classes
  - Mouse/keyboard event handling
  - Real-time visual feedback
- Backend: nodes/matte/roto_bezier.py
  - DetonateRotoBezier, DetonateRotoBezierFromImage nodes
  - High-quality rasterization pipeline
  - Anti-aliasing via supersampling
  - Distance transform-based feathering
- Math: utils/bezier_utils.py
  - de Casteljau recursive evaluation
  - Spline discretization (uniform & adaptive)
  - Tangent computation for normals

Why This Matters:
- #1 requested feature by compositors
- Replaces Natron/Nuke RotoPaint for basic rotoscoping
- Essential for garbage mattes and subject isolation
- Best-in-class implementation with modern web UI

Documentation:
- Comprehensive NODE_SPECIFICATIONS.md entry (Tier 5)
- Updated README with feature highlights
- Detailed technical specifications and workflow examples

Total Nodes: 36 (9 Tier 1 + 8 Tier 2 + 1 Cryptomatte + 8 Tier 3 + 8 Tier 4 + 2 Tier 5)
Version: 0.7.0
 2025-11-22 18:23:38 +00:00
Claude 20df905356 feat: Phase 1 foundation setup - project infrastructure complete 
This commit establishes the complete foundation for ComfyUI_Detonate:

## Project Structure
- Created organized directory structure for all node categories
- Set up proper Python package structure with __init__.py files

## Package Configuration
- Added pyproject.toml with project metadata and dependencies
- Created requirements.txt for core dependencies
- Added requirements-dev.txt for development tools

## Documentation
- Created comprehensive NODE_SPECIFICATIONS.md documenting all Priority 1 nodes
- Includes exact formulas, parameters, and behaviors from Nuke/Fusion
- Documents: Merge, Transform, Blur, Grade, Premult/Unpremult, Shuffle, ColorCorrect, Erode/Dilate
- All specifications backed by official Nuke and Fusion documentation

## Core Utilities
- tensor_utils.py: Tensor validation, alpha channel handling, resizing
- image_processing.py: Premultiply/unpremultiply, luminance, masking, blending
- color_math.py: RGB/HSV conversion, saturation, contrast, gamma, lift/gain/offset

## Node Registration
- Created __init__.py with ComfyUI node registration system
- Ready to register nodes as they are implemented

All utility modules are fully documented with docstrings and type hints.
Ready to begin Priority 1 node implementation.
 2025-11-22 05:49:08 +00:00