Settings

Configure the Editor platform

Thumbnail Interval (seconds)

Seconds between sprite sheet frames

HLS Bitrates

Comma-separated bitrate ladder

FFmpeg Threads

X-CLIP Frame Sample Rate

Frames sampled per 5-min segment for scene classification

Min Face Size (px)

Minimum detectable face dimension

Face Confidence Threshold

Min detection score (0.0–1.0)

Voice↔Face Overlap Threshold

Min temporal overlap % to link a speaker to a face track (higher = stricter)

Default Format

CRF (Quality)

Lower = better quality (18–23 recommended for MP4)

Max Upload Size

Maximum single file upload size

Data Root Path

Removes intermediate HLS quality variants, keeping only 720p and 1080p

OpenRouter API Key

Required for LLM-powered segment analysis and content insights

HuggingFace Token (pyannote)

HuggingFace token for speaker diarization models (must accept model terms)