Fill a timeline gap with generated B-roll through an API

To fill a gap in an edit with generated B-roll through an API, generate a short clip with POST /v1/videos, then reference its media.sume.com URL as a slot in a Timeline 1.0 render. The two calls are separate jobs, so you choose the clip, check it, and only then pay for the final cut. Adobe's September 2026 Premiere release added a Generative Media Tool that creates video and sound effects from text inside the timeline, which is the workflow people are now searching for outside Adobe's editor.

What Adobe announced and what Sume does

Adobe's release notes describe generating clips from text in the timeline, for example a 3-second transition. Sume has no timeline editor UI for this; it exposes the same two steps as API calls. The facts below are dated.

Steps

1. Generate the clip. Discover current models with GET /v1/videos/models rather than hard-coding one, and keep the request short: a gap filler is a few seconds.

2. When the job finishes, take the result's video URL. It is a durable media.sume.com artifact, which is exactly what Timeline 1.0 requires: every source URL must already be your workspace's artifact or asset.

3. Call /v1/timeline-1.0/plan with the full slot list. It validates the program and returns billable_minutes and estimated_cost_usd_micros without creating a job.

4. Render with an Idempotency-Key.

Place the generated clip as its own slot. Use fit: "cover" if the aspect ratio differs, and a short fade or dissolve transition (at most 1 s) so the join does not read as a cut to a different world.

{
  "audio": {"url": "https://media.sume.com/artifacts/artf_demo/voice.wav", "duration_seconds": 20},
  "video": [
    {"source_url": "https://media.sume.com/artifacts/artf_demo/talk.mp4", "start": 0, "duration": 12},
    {"source_url": "https://media.sume.com/artifacts/artf_demo/generated-broll.mp4",
     "start": 12, "duration": 4, "fit": "cover", "transition": {"type": "dissolve", "duration": 0.5}},
    {"source_url": "https://media.sume.com/artifacts/artf_demo/talk.mp4", "source_in": 12, "start": 16, "duration": 4}
  ]
}

Choosing the fit

fit decides how a clip that does not match the frame is handled: cover crops to fill, contain letterboxes, stretch distorts and blur fills the edges with a blurred copy. For generated B-roll, cover is usually right because you control the prompt and the aspect ratio can be chosen up front; use blur only for a source you cannot reframe.

Limits

A slot must be at least 0.2 s, the first slot must start at 0, and a render holds 1 to 200 slots with audio up to 1800 s. The generated clip is only as long as the model allows, so a 4-second gap is easy and a 20-second gap needs several clips. Generation is not deterministic: the same prompt can give a different clip, so review each one before it goes into the render. Sume does not generate matching sound effects inside Timeline; supply the audio track you want. Sume's render is ffmpeg on a worker with no inference, so its price is the timeline rate of $0.10 per output minute, separate from the generation cost. See also AI B-roll generator.