Loki Studio v1.0.0

← All updates

NEW FEATURES

### Core Workflow

• Dashboard with video profile management
• Video collection and organization by game/series
• Pipeline status tracking (transcribed, metadata, thumbnail, uploaded)

### Transcription (Muninn)

• GPU-accelerated Whisper transcription via CTranslate2
• Multiple model sizes: tiny, base, small, medium, large-v3, large-v3-turbo
• Voice Activity Detection (VAD) for accurate speech boundaries
• Word-level timestamps for caption generation
• Support for 99+ languages with auto-detection

### Multi-Language Translation

• NLLB-200 neural translation for 34+ languages
• Word-level timing preservation during translation
• Thread-safe TranslationWorker runs in background, non-blocking UI

### Metadata Generation

• AI-powered title, description, and tag generation
• Multiple LLM providers: Claude, GPT-4, LM Studio, Ollama
• AI Personality System - personalities affect all content generation
• GBNF grammar constraints for structured JSON output

### Thumbnail Creation

• Frame extraction from video at any timestamp
• Text overlay with customizable fonts, colors, and positions
• Image compositing with background images
• Batch thumbnail generation

### Timeline Editor

• Multi-track video timeline with waveform display
• Clip extraction with frame-accurate cuts
• Caption preview and burn-in export
• EDL import from Adobe Premiere Pro
• AutoCut feature for automatic cut word detection

### Caption System

• Three caption styles: One Word Pop, Sliding Window, Full Segment
• Real-time caption preview during playback
• ASS/SRT export for external use
• Customizable fonts, colors, positioning, and timing

### YouTube Upload

• OAuth 2.0 authentication with YouTube Data API
• Batch upload with scheduling support
• Privacy settings (public, unlisted, private)
• Playlist management and creation
• Made for Kids compliance setting

### Settings & Configuration

• Power User Settings for YouTube defaults
• Application Settings for LLM and transcription config
• Profile Settings for per-game/series configuration
• Searchable Help documentation

TECHNICAL HIGHLIGHTS

• Qt 6.10 with full QML UI (migrated from Qt Designer)
• Norse-themed modular architecture (Muninn, Huginn, Freyja, Mimir, etc.)
• GStreamer-based video playback engine
• FFmpeg for video processing (statically linked in DLLs)
• Thread-safe design with Qt::QueuedConnection for all worker signals

LICENSING

• Free tier with core functionality
• Pro tier via Gumroad license verification
• Developer Mode (forseti_dev.dll) for testing

SYSTEM REQUIREMENTS

• Windows 10/11 64-bit
• NVIDIA GPU recommended for Whisper acceleration (CUDA 12)
• 8 GB RAM minimum, 16 GB recommended
• 2 GB disk space + model storage

KNOWN ISSUES

• First transcription after model load takes a few seconds to initialize
• Large video files may take time to generate waveforms
• Some antivirus software may flag unsigned executables

← All updates  ·  Source-of-truth changelog on GitHub