Video editing and production
The video commands work with existing clips and content-production jobs.
Use AI video generation when you want a video model
to generate a scene from a prompt or reference image.
1. Prepare your inputs
Use reachable video URLs for editing operations. For local recordings, upload the file and wait for its processed URL. Keep the source asset IDs so you can retrieve fresh URLs later.
2. Edit a clip
Convert a clip to MP4:
Remove its audio or reverse playback:
Change playback speed with --playbackrate. A value of 2 doubles the speed;
0.5 slows it to half speed:
These are separate operations. To chain them, use the result URL from one command as the input for the next.
3. Merge clips
Provide at least two URLs in the intended order:
Inspect the combined video for pacing, transitions between source clips, and whether their dimensions suit the destination channel.
4. Add B-roll
The command also accepts an existing asset identifier through --asset. Choose
the source form that matches your workflow. Review the generated B-roll for
relevance to the narration before publishing.
5. Produce narrated content
video:script-to-video and video:text-to-video expose production inputs such as
a title, description, keywords, voice, tone, language, and output format:
Use --voice-id, --logo-id, and --caption-style-id when you already have the
appropriate identifiers. Version 1.8.0 does not expose a script-file input flag;
check each command’s help before preparing an automated production job.
6. Check progress and output
Without --wait, save the returned task_id:
Standard video editing waits up to 300 seconds. Narrated production can wait up
to 300 seconds for the task and another 300 seconds for an export when the task
returns an export ID. --should-export is enabled by default for the two narrated
production commands.
A waiting timeout does not cancel the remote job. Inspect the existing task before retrying. Review the final video and use its output URL in your publishing workflow once it is ready.
