Scene to Video Pipeline

Add scenes (generate via Wan2.2, or drop in a premade clip), write the narration, hit submit.

Narration Script

Full combined narration, voiced via ElevenLabs.

ElevenLabs API Key

Leave blank to use the key from server .env.

Speed / Quality

Test mode uses fewer sampling steps for a quick, low-quality draft — use it to check prompts/motion before a full-quality run. (The Wan2.2 ti2v-5B model only supports 704*1280 / 1280*704 — same pixel count either way, so orientation doesn't affect speed; steps is the only speed lever.)

Test mode (fast draft, low quality) Orientation

Idle.

Faceless Video

Images + Ken Burns pan/zoom + narration + captions. Runs fully local — macOS say for voice (free, offline), no GPU boxes, no paid API.

Image(s) — shown in order, duration split evenly across narration length

Narration Script

Voiced locally via macOS say. Also split into captions sentence-by-sentence.

Resolution

Caption Font

Server Platform

Where this job will actually run. Picks the right local TTS engine + fonts — macOS uses say + system fonts, Linux uses espeak-ng + bundled Liberation/DejaVu fonts (both free, offline).

Idle.

Direct Lip-Sync

Upload a video and a voice track — MuseTalk lip-syncs them directly, no generation/TTS.

Video

Drop video here or click to browse

Voice / Audio

Drop audio here or click to browse

Idle.

Admin

Box health, disk usage, running jobs.

Protected