Input & Scene Planning
Progress and Cancellation
Understanding the generation stages: transcription, research, scene planning, script, voice, and rendering.
2 min read•Last updated: September 2026
Generation Pipeline Stages
During video generation, the progress panel tracks each phase in real time:
- Transcribing audio → English: Transcribes spoken audio and establishes sentence timestamps.
- Web search: Fetches relevant real-time technical context (if enabled).
- Planning: The AI director structures logical scene boundaries.
- Script: Narration and visual element definitions are generated.
- Voice: Narration audio is synthesized or aligned.
- Rendering: Frames are drawn and encoded into MP4.
Do Not Close the Tab During Render
During video decoding and MP4 finalization, avoid switching tabs or closing the browser window. Canceling after generation has started may consume a credit because upstream computing has already completed.