Multi-aspect reframing
Three deliverables from one timeline — 9:16, 1:1, 16:9 — with subject tracking that keeps faces safe-zone aware.
A focused toolkit, not a kitchen sink. Each module exists because manual versions waste hours per reel.
Drop in an unedited recording. The model finds the hook, drops filler words and silences, tightens pacing, and proposes a cut shaped to platform conventions.
A scene-aware LUT engine that reads each shot — skin, sky, foliage — and applies a look without crushing detail. Twelve built-in styles, plus your own.
Auto-ducking, noise reduction, room-tone matching across cuts, and platform-correct loudness targets. Score from a license-cleared library, or bring your own.
Transcripts accurate to 99% on clear English audio. Style per-word, animate on emphasis, export burned-in or as a sidecar SRT.
Three deliverables from one timeline — 9:16, 1:1, 16:9 — with subject tracking that keeps faces safe-zone aware.
Semantic search against your library — surfaces relevant b-roll while you cut, instead of after.
Queue and publish directly to TikTok, Instagram, YouTube Shorts, X and LinkedIn. Time-zone aware.
Inpainting model trained on common social watermarks. Use only on assets you own — we check.
Fonts, colors, LUTs, caption styles, intro/outro stings — locked across the workspace.
Role-based access. Junior editors propose cuts, leads approve. Full version history per asset.