Face swap beta, 9-second clip: you reserve the 15-second price

Sume's Avatar Face Swap beta reserves the price of the maximum 15-second source clip, even when your source is 9 seconds. The reservation is $2.76 at standard (15 x $0.184), $3.675 at plus (15 x $0.245) and $8.25 at max (15 x $0.55). Budget with those figures, not with 9 seconds x the rate.

What the catalog says

The catalog rows for sume/avatar-face-swap/v1.0 describe each tier as reserving with the matching Avatar Video public pricing for the maximum beta source-video duration. They add that final workflow pricing may be refined before general availability. Because it is a beta, the safe reading is: plan for the reserved figure per job and treat anything lower as a bonus you confirm on the job, not a number to build a budget on.

Because the figure is a reservation for the beta, it is wise to run one clip first. Read the job result, note the amount captured, and compare it with the reserved figure. The catalog says final workflow pricing may be refined before general availability, so a one-clip test tells you more than a model of the charge.

Source clip rules that decide the job

The docs set the source clip at roughly 4 to 15 seconds with usable audio, a fetchable public HTTPS URL, and no signed, private or provider task URLs. The request takes avatar_handle, video_url and quality only. There is no default for quality in the beta; omit it and the endpoint does not guess.

A 9-second clip is well inside the window, so you gain nothing in eligibility by padding. What you can do is batch: if you have four 9-second sources, they are four jobs, each reserving the 15-second price.

Source audio matters. The docs require usable audio in the 4 to 15 second source, and the endpoint takes no transcript or prompt, so there is nothing to supply in the request to change what is said. If the clip has no speech, pick a different source or use the talking-video route, which is driven by a script.

• Required fields: avatar_handle, video_url, quality.
• No prompt, transcript, duration or aspect-ratio fields.
• Poll resource_status for readiness and job_status for the job.

A budget for twelve clips

Twelve 9-second source clips at standard reserve 12 x $2.76 = $33.12. At plus the same twelve reserve 12 x $3.675 = $44.10. If your funding check is stricter than the final charge, you need that balance available at submit time, so top up to the reserved total rather than to the smaller per-second estimate.

Put the reservation in your planning sheet as its own column, separate from any per-second estimate. For a campaign of 30 sources at plus the reserved total is 30 x $3.675 = $110.25, and at standard it is 30 x $2.76 = $82.80. Submit in waves of five or ten so the balance you must hold at once stays small, and read each finished job before you submit the next wave. That way a pricing change in the beta shows up in the first wave and not in the last.