Recipes

Each recipe is what you would say to an assistant connected to PageToVid, followed by the tool calls it makes. You never have to write the calls yourself — they are shown so you know what to expect and what to ask for.

  1. A 30-second ad for a website
  2. Fix one scene without starting over
  3. Correct one figure
  4. A film from data, with no website
  5. A presenter walking through your product
  6. Generated clips, a model per scene
  7. Re-time captions only
  8. Recurring videos from a feed

A 30-second ad for a website

Make a 30-second vertical ad for stripe.com, for founders. Lead with “Get paid in minutes”.

  1. get_account — plan, credits, what is available.
  2. inspect_page url: https://stripe.com aspect_ratio: 9:16 — what the camera can point at, in the phone layout.
  3. create_video url · goal: ad · aspect_ratio: 9:16 · target_seconds: 30 · focus: "Audience: founders. \"Get paid in minutes\"".
  4. get_video until done, then read warnings.
  5. get_storyboard — check each shot frames something different and at least one does something.

Fix one scene without starting over

Scene 3 shows the footer while the voice talks about pricing — point it at the pricing table.

  1. get_storyboard — find scene 3’s scene_id.
  2. inspect_page — find the pricing table’s selector.
  3. update_storyboard set_shot scene_id selector action: spotlight.
  4. rerender_video — 40 credits (a render); only what changed is filmed again.

Correct one figure

The chart says 40% but it should be 45%.

get_storyboard → update_storyboard set_data with the scene’s whole data, figure corrected → rerender_video.

A film from data, with no website

Turn these quarterly numbers into a 20-second chart video.

list_motions → create_animation with scenes you write: a chart with the points, a bignum for the headline figure, a titlecard to close. Each narrated scene lasts as long as its words.

A presenter walking through your product

Put a presenter in the corner who explains how to use the site, like a YouTuber. No narrator.

create_character → create_video → update_storyboard with set_shot (different, active shots), rewrite_narration "" and set_inset (the line in quotation marks) per scene → rerender_video. The full walk-through is in A presenter in the corner.

Generated clips, a model per scene

Make the opening shot cinematic with Seedance, and keep the rest filmed from the site.

update_storyboard set_visual image with data: { clip: true, model: "seedance-2.0-mini", generate: "…" } → rerender_video. See AI models.

Re-time captions only

The subtitles are a little behind the voice.

rerender_video captions_only: true — no new take, free once per render.

Recurring videos from a feed

Every time I publish a blog post, make a 30-second video of it.

create_format (the shape of every episode) → validate_format (free) → create_show with the feed → preview_episode (free) → the show runs on its schedule, never renders the same item twice, and holds its first episode until you confirm it.