Help center how-to video with AI: avatar or screenshots

To make a help-center how-to video with AI, write the steps as about 75 spoken words, then pick one of two routes on Sume. Route A is a talking avatar with the steps as its script: a 30-second clip is $5.52 on Standard, $7.35 on Plus or $16.50 on Max, from the per-second rates on the API pricing page. Route B keeps your real screenshots on screen, adds a text-to-speech voice, and joins them with a Timeline render, for roughly 12 cents a clip.

Support teams usually need the second route. A how-to is about where to click, and a generated background cannot show your product's real buttons. The avatar route fits the opener of a help center, a welcome clip or a policy change, where a face reads as a person talking.

Which route fits which kind of support article?

Use the avatar for announcements and for articles with no interface in them: shipping changes, refund policy, a new support-hours notice. Use screenshots and a voice for anything with steps in your product. You can also combine them: a 6-second avatar greeting first, then the screenshot walkthrough.

How do I build the screenshot route?

Write one sentence per step and keep each under about 15 words, because a voice reads roughly 150 words a minute. For 30 seconds that is around 450 characters of script, which is $0.0214 of TTS at $0.0475 per 1,000 characters.

Import each screenshot to Sume first. Timeline 1.0 only accepts files that already live in your workspace's media.sume.com storage, so use a media import for the screenshots and the voice file. Then give the render one audio spine (the voice) and one slot per screenshot with a start and a duration. Stills are held static, and a fade transition of a quarter second between steps reads as a step change without a jarring cut.

Before you spend anything, call the plan endpoint. It compiles your document and returns the duration, the segment count and an estimated cost without creating a job.

What does a ten-article help center cost?

Ten 30-second clips on the screenshot route is about $1.21: ten voices at $0.0214 and ten renders at $0.10, since each render bills a rounded-up output minute. The same ten on avatar Plus is $73.50. If a policy changes, the screenshot route is also the cheap one to redo: edit one sentence, regenerate the voice, render again.

Should the clip have captions?

Yes, for support. Many viewers watch with the sound off while at work. On the avatar route, set captions in the request and Sume burns them into the final MP4 with no separate billed caption job. On the screenshot route, send authored cues (text, start, end) to the standalone video-captions endpoint, which handles overlay text without speech recognition. The pricing page does not list a caption rate, so check the playground or dry-run before a bulk run.

Keep one how-to to one task. A viewer who needs to reset a password does not want the billing tour, and a short clip is also easier to keep current.