Square YouTube Short from vertical AI clips: 1080x1080 on Timeline

To make a square Short, keep generating 9:16 clips and set output.width and output.height to 1080 on the Timeline 1.0 render; no model needs a square mode. YouTube's help page lists square and vertical as Shorts shapes and three minutes as the upper length (read 2026-10-08). A full 180-second square Short from Wan 3.0 at 720p costs $22.80 before voice.

The only change from a vertical render

Timeline's default output is 1080 by 1920. Output width and height accept even integers from 256 to 2160, so 1080 by 1080 is a valid override. Each slot's fit decides how a 9:16 frame sits in the square: cover (the default) crops, contain letterboxes, stretch distorts, blur keeps the whole frame over a blurred fill.

{
  "audio": { "mode": "silence", "duration_seconds": 60 },
  "output": { "width": 1080, "height": 1080, "fps": 30 },
  "video": [
    { "source_url": "https://media.sume.com/artifacts/artf_demo/a.mp4",
      "start": 0, "duration": 30, "fit": "cover" },
    { "source_url": "https://media.sume.com/artifacts/artf_demo/b.mp4",
      "start": 30, "duration": 30, "fit": "blur" }
  ]
}

Pick the fit before you generate

A 9:16 frame cropped to a square keeps the middle 56 percent of the height, so prompt for the subject centered. If the shot has important top and bottom framing, use blur.

Silence mode declares a length without a spine file. If the Short needs sound, send a voice or music file as audio.url instead of silence mode; soundtrack.duck_db needs a real spine, not silence.

Cost by length

Clips are Wan 3.0 at 30 seconds each; the render is $0.10 per started minute.

Check before paying

POST /v1/timeline-1.0/plan compiles the document without a job or a reserve and returns billable_minutes and an estimate, so a wrong dimension costs nothing.

Framing a square from a vertical clip

A 1080 by 1080 canvas shows a centered square window of a 1080 by 1920 frame under cover fit, so about 44 percent of the vertical height is cropped away. Write prompts with the subject and any on-screen action in the middle band, and avoid putting key objects at the very top or bottom of the frame. If you need the whole picture, use contain or blur and accept bars or a soft fill. To choose, render a 10-second test of one clip in each fit (the render costs $0.10 for the first started minute) and compare before committing a full 180 seconds.