Mood still plus Music 1.0: 13.5 cents a scene, 8 scenes for $1.08

A mood still generated at low quality costs 1 cent and a Music 1.0 track costs $0.125, so one image-conditioned scene is 13.5 cents on Sume. Eight scenes are 8 x $0.135 = $1.08. Music 1.0 does not charge extra for image conditioning: the price is a flat $0.125 whether or not you send image_url.

How the pieces fit

Music 1.0 takes a text prompt and an optional image. The image is a public HTTPS URL in image_url; send null to clear it. You can use a still you already have, an accepted frame from a video, or one you generate. Generating one with gpt-image-2.5 at quality low is the cheap route: the Sume catalog bills a 1024-class low-quality image at 1 cent.

The image conditions the music, but the docs call the seven brief axes (emotion, genre, tempo, key, instruments, arc, era) creative directions, not guaranteed values. The still adds mood; the text still carries tempo and instrumentation.

One consistent score vs contrast between scenes

For scenes in one project that should contrast, the docs advise changing the broad genre family, the tempo by at least 12 BPM, and the lead instrument. If you want one consistent score, keep the brief the same and pass the accepted scene still as image_url when it applies. Eight scenes with the same tempo, key and lead instrument will sound like one piece; eight scenes with 12 BPM steps will not.

Because there is no seed, plan one extra take per scene for a decision: at 12.5 cents a track, a second take on all eight scenes adds $1.00.

After a policy rejection

If a request is rejected on policy grounds, change the flagged content but keep the musical brief; do not reduce the prompt to a generic bed. Retry only inside the budget you were authorized to spend. The rejection is on the content, so swapping the still is often enough.

Using it in a real project

Run the stills first, pick the ones you like, and only then spend on music. Eight low-quality stills cost 8 cents in total, so you can generate three per scene (24 cents) and still be under a quarter for the whole set. Music is the larger line, at $1.00 for eight tracks, so choose which stills deserve a track.

The image must be a public HTTPS URL. Use a durable HTTPS URL so the image is still reachable when the job runs.