13 — Job Queue

13 — Job Queue

The Job Queue tracks all AI generation tasks — images, videos, talking heads, audio, and voice synthesis — across all features of the app. Jobs run in the background, so you can continue editing while generation completes.


How It Works

When you click Generate on any AI feature, the request is sent to the Replicate API (or the relevant provider) and a job entry is created in the queue.

The app polls the provider every 3–5 seconds until the job finishes or fails. Jobs continue tracking even if you switch panels or navigate to a different section.

Job states:

State Meaning
Submitted Job sent to provider; waiting to start
Running Provider is actively generating
Succeeded Generation complete; result is ready
Failed Generation encountered an error

Accessing the Queue

The Job Queue panel is accessible within the AI Generate section. It displays:
– All currently running and recently completed jobs
– Job type (Image, Video, Talking Head, Audio)
– Current status
– Elapsed time
– A thumbnail or audio player for completed results


Recent Results

The six most recently completed jobs appear in the Recent Results section as cards. Each card shows:
– A preview (image thumbnail, video thumbnail, or audio player)
– Download button
Add to Timeline button


Persistence

Job queue data is stored locally in your browser. This means:
– Jobs survive navigating away from a panel
– Jobs survive app reload
– You can start a long generation job, close the AI Generate panel, do other editing work, and come back to find your results

Jobs are retained until you clear them or the app’s local storage is cleared.


Adding Results to the Timeline

For any completed job:
Images — click Add to Timeline to insert as a still-frame video segment
Videos — click Add to Timeline to insert as a video segment at the current playhead position
Audio — click Add to Timeline to insert into the audio track at the current playhead position


Handling Failed Jobs

If a job shows Failed, click the job entry to see the error message from the provider. Common causes:
– Invalid or expired API key
– Model temporarily unavailable (retry after a minute)
– Input that violates the provider’s content policy
– Cloud storage URL not accessible (for talking head / reference image jobs)

To retry, adjust your inputs if needed and click Generate again.


Previous: Long-Form Video Pipeline | Next: Settings Reference →