VideoTools/TODO.md

VideoTools TODO (v0.1.0-dev22+ plan)

This file tracks upcoming features, improvements, and known issues.

Documentation: Address Platform Gaps

Priority: High

• Create Native Windows Guide:
  • Create a comprehensive installation and usage guide for native Windows.
  • This guide should be on par with the existing Linux guide.
  • Refactor INSTALLATION.md to be a central hub linking to platform-specific instructions.

Critical Priority: dev22

VIDEO PLAYER IMPLEMENTATION

CRITICAL BLOCKER: All advanced features (enhancement, trim, advanced filters) depend on stable player foundation.

Current Player Issues (from PLAYER_PERFORMANCE_ISSUES.md):

1. Separate A/V Processes (lines 10184-10185 in main.go)

   • Video and audio run in completely separate FFmpeg processes
   • No synchronization mechanism between them
   • They will inevitably drift apart, causing A/V desync and stuttering
   • FIX: Implement unified FFmpeg process with multiplexed output

2. Audio Buffer Too Small (lines 8960, 9274 in main.go)

   • Currently 8192 samples = 170ms buffer
   • Modern systems need 100-200ms buffers for smooth playback
   • FIX: Increase to 16384-32768 samples (340-680ms)

3. Volume Processing in Hot Path (lines 9294-9318 in main.go)

   • Processes volume on EVERY audio sample in real-time
   • CPU-intensive and blocks audio read loop
   • FIX: Move volume processing to FFmpeg filters

4. Video Frame Pacing Issues (lines 9200-9203 in main.go)

   • time.Sleep() is not precise, cumulative timing errors
   • No correction mechanism if we fall behind
   • FIX: Implement adaptive timing with drift correction

5. UI Thread Blocking (lines 9207-9215 in main.go)

   • Frame updates queue up if UI thread is busy
   • No frame dropping mechanism
   • FIX: Implement proper frame buffer management

6. No Frame-Accurate Seeking (lines 10018-10028 in main.go)

   • Seeking kills and restarts both FFmpeg processes
   • 100-500ms gap during seek operations
   • No keyframe awareness
   • FIX: Implement frame-level seeking without process restart

Player Implementation Plan:

Phase 1: Foundation (Week 1-2)

• Unified FFmpeg Architecture

  • Single process with multiplexed A/V output using pipes
  • Master clock reference for synchronization
  • PTS-based drift correction mechanisms
  • Ring buffers for audio and video

• Hardware Acceleration Integration

  • Auto-detect available backends (CUDA, VA-API, VideoToolbox)
  • FFmpeg hardware acceleration through native flags
  • Fallback to software acceleration when hardware unavailable

• Frame Extraction System

  • Frame extraction without restarting playback
  • Keyframe detection and indexing
  • Frame buffer pooling to reduce GC pressure

Phase 2: Core Features (Week 3-4)

• Frame-Accurate Seeking

  • Seek to specific frames without restarts
  • Keyframe-aware seeking for performance
  • Frame extraction at seek points for preview

• Chapter System Integration

  • Port scene detection from Author module
  • Manual chapter support with keyframing
  • Chapter navigation (next/previous)
  • Chapter display in UI

• Performance Optimization

  • Adaptive frame timing with drift correction
  • Frame dropping when UI thread can't keep up
  • Memory pool management for frame buffers
  • CPU usage optimization

Phase 3: Advanced Features (Week 5-6)

• Preview System

  • Real-time frame extraction
  • Thumbnail generation from keyframes
  • Frame buffer caching for previews

• Error Recovery

  • Graceful failure handling
  • Resume capability after crashes
  • Smart fallback mechanisms

ENHANCEMENT MODULE FOUNDATION

DEPENDS ON PLAYER COMPLETION

Current State:

• Basic filters module with color correction, sharpening, transforms
• Stylistic effects (8mm, 16mm, B&W Film, Silent Film, VHS, Webcam)
• AI upscaling with Real-ESRGAN integration
• Basic AI model management
• No content-aware processing
• No multi-pass enhancement pipeline
• No before/after preview system

Enhancement Module Plan:

Phase 1: Architecture (Week 1-2 - POST PLAYER)

• Model Registry System

  • Abstract AI model interface for easy extension
  • Dynamic model discovery and registration
  • Model requirements validation
  • Configuration management for different model types

• Content Detection Pipeline

  • Automatic content type detection (general/anime/film)
  • Quality assessment algorithms
  • Progressive vs interlaced detection
  • Artifact analysis (compression noise, film grain)

• Unified Enhancement Workflow

  • Combine Filters + Upscale into single module
  • Content-aware model selection logic
  • Multi-pass processing framework
  • Quality preservation controls

Phase 2: Model Integration (Week 3-4)

• Open-Source AI Model Expansion

  • BasicVSR integration (video-specific super-resolution)
  • RIFE models for frame interpolation
  • Real-CUGan for anime/cartoon enhancement
  • Model selection based on content type

• Advanced Processing Features

  • Sequential model application capabilities
  • Custom enhancement pipeline creation
  • Parameter fine-tuning for different models
  • Quality vs Speed presets

TRIM MODULE ENHANCEMENT

DEPENDS ON PLAYER COMPLETION

Current State:

• Basic planning completed
• No timeline interface
• No frame-accurate cutting
• No chapter integration from Author module

Trim Module Plan:

Phase 1: Foundation (Week 1-2 - POST PLAYER)

• Timeline Interface

  • Frame-accurate timeline visualization
  • Zoom capabilities for precise editing
  • Scrubbing with real-time preview
  • Time/frame dual display modes

• Chapter Integration

  • Import scene detection from Author module
  • Manual chapter marker creation
  • Chapter navigation controls
  • Visual chapter markers on timeline

• Frame-Accurate Cutting

  • Exact frame selection for in/out points
  • Preview before/after trim points
  • Multiple segment trimming support

Phase 2: Advanced Features (Week 3-4)

• Smart Export System
  • Lossless vs re-encode decision logic
  • Format preservation when possible
  • Quality-aware encoding settings
  • Batch trimming operations

DOCUMENTATION UPDATES

• Create PLAYER_MODULE.md - Comprehensive player architecture documentation
• Update MODULES.md - Player and enhancement integration details
• Update ROADMAP.md - Player-first development strategy
• Create enhancement integration guide - How modules work together
• API documentation - Player interface for module developers

Future Enhancements (dev23+)

AI Model Expansion

• Diffusion-based models - SeedVR2, SVFR integration
• Advanced restoration - Scratch repair, dust removal, color fading
• Face enhancement - GFPGAN integration for portrait content
• Specialized models - Content-specific models (sports, archival, etc.)

Professional Features

• Batch enhancement queue - Process multiple videos with enhancement pipeline
• Hardware optimization - Multi-GPU support, memory management
• Export system - Professional format support (ProRes, DNxHD, etc.)
• Plugin architecture - Extensible system for community contributions

Integration Improvements

• Module communication - Seamless data flow between modules
• Unified settings - Shared configuration across modules
• Performance monitoring - Resource usage tracking and optimization
• Cross-platform testing - Linux, Windows, macOS parity

Technical Debt Addressed

Player Architecture

• Identified root causes of instability
• Planned Go-based unified solution
• Hardware acceleration strategy defined
• Frame-accurate seeking approach designed

Enhancement Strategy

• Open-source model ecosystem researched
• Scalable architecture designed
• Content-aware processing planned
• Future-proof model integration system

Notes

• Player stability is BLOCKER: Cannot proceed with enhancement features until player is stable
• Go implementation preferred: Maintains single codebase, excellent testing ecosystem
• Open-source focus: No commercial dependencies, community-driven model ecosystem
• Modular design: Each enhancement system can be developed and tested independently