Tab Reference Guide

Config Status

Shows the currently loaded configuration file and its path

Quick Stats

• Videos in upload folder
• Transcripts completed
• Metadata generated
• Thumbnails created
• Videos uploaded

Recent Activity

Last 10 actions taken, useful for tracking workflow progress

Quick Actions

• Open root directory
• Open upload folder
• Open config file
• Refresh all stats

• Check stats regularly to monitor workflow progress
• Recent Activity helps identify bottlenecks
• Keep config file backed up

Video List

Shows all videos in toupload/ folder

Columns: Name, Duration, Status, Transcript

Settings Panel

• Model: Dropdown shows only installed models (large-v3-turbo recommended)
• Device: CUDA (GPU) for 10-20x faster, or CPU
• Compute Type: float16 (GPU), float32, int8 (CPU)
• VAD Mode: Auto, Silero, Energy, or Off
• Language: 99+ languages with auto-detect
• Translation: Any-to-any translation between 30+ languages (NLLB-200)
• Speaker Diarization: Identify up to 10 speakers
• Word Timestamps: Essential for karaoke captions

Progress Display

• Current video progress bar
• Overall progress (X/Y videos)
• ETA (estimated time remaining)
• GPU utilization (if available)

Log Window

• Detailed processing logs with timestamps
• Error messages and warnings
• Speed metrics and performance data

• Use large-v3-turbo - Best balance of speed and quality
• Enable Word Timestamps if you want karaoke-style captions
• Use GPU (CUDA) - 10-20x faster than CPU
• Enable Silero VAD for intelligent silence removal
• Performance: ~29x realtime on RTX 4090 with large-v3
• Cost: $0.00 - completely free, runs locally

Video List

Shows videos with transcripts. Columns: Name, Status, Title, Tags

Preview Panel

• Title: Full title text (under 100 characters)
• Description: Full description with line breaks and keywords
• Tags: Comma-separated tag list for SEO
• Key Moments: Auto-generate YouTube chapters
• Character Counts: Shows length vs. YouTube limits

AI Provider Options

• Ollama (FREE): One-click model install from Settings
• LM Studio (FREE): Local LLM with nice UI
• OpenAI: GPT-4o-mini (~$0.01-0.05/video)
• Anthropic: Claude Sonnet (~$0.02-0.10/video)

AI Personalities (37+)

• Dad Joke, Space Marine, Product Reveal
• Professional, Casual, Enthusiastic, Mysterious
• Custom personalities available per profile

Templates

• Description Header:Prepended to all descriptions
• Description Footer:Appended to all descriptions
• Custom Prompts:Override AI behavior (advanced)

• Specific config = better metadata:Include target_audience and genre in your config
• Cloud providers usually superior:OpenAI typically better quality than local AI
• Adjust temperature:0.9 for creative, 0.5 for conservative titles
• Always review:AI isn't perfect - manually edit if needed
• Save templates:Reuse good metadata for similar videos

Example Generated Metadata:

Title:

Skyrim VR - Epic Dragon Battle at College of Winterhold! (Mage Ep. 5)

Tags:

Skyrim vr, skyrim gameplay, elder scrolls, vr gaming, dragon battle, mage build, college of winterhold, destruction magic, vr rpg, skyrim playthrough

Video List

Shows videos ready for thumbnail generation. Columns: Name, Status, Thumbnail preview

Preview Panel

• Large preview (1920x1080 resolution)
• Zoom controls for detailed inspection
• Open in external viewer for full-screen display
• Regenerate button to try different frame

Logo Settings

• Logo Path:Path to logo file
• Position:Top-left, top-right, top-center, bottom-left, etc.
• Size:Max width/height (preserves aspect ratio)

Text Settings

• Title Text:Game/content name (large, white by default)
• Subtitle:Episode/description (smaller, colored)
• Font Scale:Control text size
• Colors:RGB customization for each text element
• Outline:Add outline for visibility on any background

Frame Selection

• Timestamp Range:Extract frame from specific seconds (e.g., 30-120s)
• Brightness Threshold:Reject dark frames automatically
• Max Retry:Number of attempts to find good frame

• Place PNG images in[root]/Flair/Folder
• Randomly selected or specify in config
• Examples: explosions, stars, character faces, badges
• Positioned viaFlair_positionSetting

• Logo placement:PlaceLogo.pngIn root for auto-detection
• Use bright, contrasting colorsFor text visibility
• Timestamp range 30-120s:Usually has better frames than opening scenes
• Add text outlineFor visibility on any background
• Test logo positions:Top-right common for gaming content
• Flair adds personality:Use decorative overlays to stand out

Thumbnail Layout Example:

┌─────────────────────────────────────────┐
│ [Logo] │ ← Logo (top-right)
│ │
│ SKYRIM VR │ ← Title (large, white)
│ Epic Dragon Battle! │ ← Subtitle (blue)
│ │
│ [Gameplay Frame] │ ← Video screenshot
│ │
└─────────────────────────────────────────┘

Video Queue Reordering (NEW!)

• Up/Down Arrows: Control upload priority
• Queue Persistence: Order saved across tab switches
• Drag-free: Reliable arrow button reordering

YouTube Channel Section

• Channel Dropdown: Select active channel
• Add Channel: Authorize new channel
• Validate Channel: Check credentials validity

Video List

Shows videos ready for upload (marked with checkmark)

Columns: Name, Size, Ready, Publish Date, Status, Reorder buttons

Resumable Uploads

• Handles network interruptions gracefully
• Resume where it left off after reconnection

Upload Settings

• Schedule:Days between uploads, release time (HH: MM), timezone
• Privacy:Public, unlisted, or private
• Playlist:Name (auto-created if doesn't exist)
• Category:Gaming, People & Blogs, Science & Tech, etc.
• Made for Kids:Yes/No (COPPA compliance)

Progress Display

• Per-video progress: percentage, speed (MB/s), ETA
• Overall progress: X/Y videos uploaded
• Quota usage: YouTube API quota consumed

State Management

• Uploaded Videos List:Track previously uploaded videos
• State File: .yt_uploader_state.jsonTracks history
• Skip Uploaded:Won't re-upload same video
• Reset State:Clear upload history (advanced)

• Upload video: 1,600 units
• Add to playlist: 50 units
• Upload thumbnail: 50 units
• Per video total:~1,700 units
• Daily limit:~6 videos per day (10,000 / 1,700)

• Always test with Dry Run first!Verify files before uploading
• Upload during off-peak hoursFor faster speeds
• Schedule publishes for optimal times(evenings, weekends)
• Use playlistsTo organize content
• Test with "unlisted"Before publishing publicly
• Monitor quota:10,000 units/day max
• Videos moved toDone/FolderAfter upload
• Can close Loki StudioAfter upload starts (background processing)

Live Caption System (NEW!)

• Caption Styles: One Word Pop (karaoke), Full Segment, Sliding Window
• Live Preview: See captions during playback
• Caption Burning: Burn captions directly into video
• Color Customization: Highlight, Pending, Spoken word colors
• Position Control: Bottom, Top, Center, corners

GStreamer Playback

• Hardware-accelerated video playback
• Multi-track audio support
• Real-time waveform display
• Frame-accurate seeking

Audio Track Management

• Auto-detect all audio tracks (Game, Mic 1, Mic 2)
• Per-track volume controls and mute
• Voice Dominant auto-balance for gaming
• LUFS analysis and normalization

Segment Marking

• Mark start and end points for segments
• Toggle between Keep/Remove modes
• Delete segment markers

Segment List

• 🟢Keep:Include in final video
• 🔴Remove:Exclude from final video
• Manual time editing
• Drag & drop reordering

Voice-Triggered Trimming

• Auto-Detect Markers:Find "start cut" / "end cut" in transcript
• Marker Settings:Customizable start/end phrases
• Audio Channel:Select channel for multi-channel videos
• Review & Adjust:Preview detected segments
• Apply:Create segments from markers

Export Options

• Output Path:Where to save final video
• Video Codec:Copy (fast) or re-encode (smaller files)
• Audio Channel:Select audio track for VR/multi-channel
• Export Final:Process and save final video
• Progress:Shows FFmpeg processing status

1. Record 2-hour gaming session
2. Say "start cut" when boring part starts, "end cut" when it ends
3. Load video in Timeline tab
4. Click "Auto-Detect Markers" button
5. Review auto-marked segments
6. Manually add/adjust segments if needed
7. Export final video (e.g., 2 hours → 45 minutes of action)

• Use voice markers during recordingFor hands-free editing
• Keep segments shortFor tighter pacing
• Export with "copy" codecFor fastest processing (no quality loss)
• Test with short videos firstTo learn workflow
• Export multiple versionsWith different segment selections

1. Directories

• Base Directory: Root folder for all video operations
• Create Missing: Auto-create folder structure (toupload/, Done/, Failed/)

2. LLM Providers

• Ollama (FREE): One-click model install!
• LM Studio (FREE): Local LLM option
• OpenAI: GPT-4o-mini ($0.01-0.05/video)
• Anthropic: Claude Sonnet ($0.02-0.10/video)
• Install Models: Click Install button for Ollama models
• Refresh: Detect newly installed models

3. Transcription

• Model: Only shows installed Whisper models
• Device: CUDA (GPU) or CPU
• VAD: Auto, Silero, Energy, Off
• Language: 99+ languages with auto-detect

4. Captions

• Style: One Word Pop, Full Segment, Sliding Window
• Colors: Highlight, Pending, Spoken word colors
• Position: Bottom, Top, Center, corners
• Font Size: 12-72 pixels
• Background: Toggle semi-transparent backdrop

5. Advanced

• Threading: Processing thread configuration
• Debug: Verbose logging options
• Cache: Clear cached data

Logging

• Log Level:Debug, Info, Warning, Error
• Log to File:Save logs to disk
• Log Path:Where to save log files
• Clear Logs:Delete old log files

Shortcuts

• Keyboard Shortcuts:Customize key bindings
• Quick Actions:Assign keys to common tasks

Advanced

• Developer Mode:Show debug information and API calls
• API Rate Limits:Throttle API calls to avoid hitting quotas
• Experimental Features:Enable beta features and early access

• Set default configTo auto-load your primary project on startup
• Adjust GPU memory limitIf you have limited VRAM
• Enable loggingFor troubleshooting issues
• Customize keyboard shortcutsFor your workflow
• Monitor cache sizeAnd clear periodically to free disk space