Music from a still: image_url conditioning, still $0.125

You can condition a Sume music generation on a picture by adding image_url, a public HTTPS image. The price does not change with it: the docs fix $0.125 per accepted generation, whatever the prompt length or whether an image is attached. The prompt is still required.

Request

Use the router endpoint. image_url is optional; send null to clear it if a client reuses request objects.

curl -X POST https://api.sume.com/v1/music-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: music-still-001" \
  -d '{
    "prompt": "Cinematic ambient underscore matching the mood of the reference still, instrumental only",
    "image_url": "https://example.com/moodboard.png"
  }'

One score or contrasting scenes

The docs give a rule for multi-scene projects. If scenes should contrast, change the broad genre family, move tempo at least 12 BPM apart and change the lead instrument. If one consistent score is wanted, keep continuity, and when it applies pass the accepted scene still as image_url.

Cost of scoring a storyboard

Eight scenes, one generation each, is $1.00. A retake of any scene is another $0.125, so keep the first pass to a few scenes and iterate on the ones that matter.

The image is a direction, not a guarantee. The docs describe the seven prompt axes as creative direction, not guaranteed output values; listen to the result before assembling.

Prompt and length rules

The prompt is 1 to 5000 characters, and exclusions go in the positive prompt. duration and duration_seconds are rejected, so control the length in the prompt itself, for example "a 2-minute track" or a timed outline such as [0:00-0:30] Intro: ....