Motion fieldnotesOpenAI film study
Interface sequences · Rendered draft

Step a lightbox through ad variants on a fixed beat while the frame resizes to each new aspect ratio

Study the source, inspect the editable build, then adapt the treatment to your own content.

01 / OpenAI source 1:08.3–1:13.3 · Silent research excerpt
02 / Remotion draft Original illustrated content · Editable code

What the source does

A lightbox shows a square ad card with a small brand kicker, a large serif headline, a date and a product photo. Every ~48 frames the card hard-swaps to the next variant: two landscape 4:3 versions (~f14, ~f62), two 16:9 versions (~f110, ~f158), a very wide banner (~f206) and a portrait (~f254). The card is always re-centred and the arrows and action bar stay put; the left text column and the window do not change.

Fixed-tempo carousel stepping: each step is a hard swap that re-fits the centred card to a different aspect ratio without any slide or fade.

Phase notes

  1. Frame 0 — Square variant
  2. Frame 14 — Landscape 4:3
  3. Frame 62 — Landscape 4:3 variant
  4. Frame 110 — Landscape 16:9
  5. Frame 158 — 16:9 variant
  6. Frame 206 — Wide banner
  7. Frame 254 — Portrait

Use this example

  1. Watch the source and draft at normal speed. Compare the order of actions, the held states, and the transition into the result.
  2. Download the project and read its README before changing inputs. The package includes its implementation and declared dependencies.
  3. Replace the sample content through the package's documented input or content builder. Keep source footage in the research comparison.
  4. Render your adaptation and inspect text fit, framing, intermediate states, and entry and exit frames.
A runnable draft is not a fidelity approval.

The source and remake are different assets. The original code-based illustration remains editable; source footage is a research reference. Read the retained findings below before using the choreography.

Recorded limits

  • Visual fidelity is an authored draft judgement from sampled source frames; no independent blind review or native Resolve execution.

Inspect the original verification record ↗