Technique / PunchCut

Cut directly from context to detail

Make an interaction readable without spending time on camera travel.

Run the original

4.8 s · 30 fps · silent

Original neutral assets · 4.8 seconds · phase values are chosen defaults, not source timings.

npm run render -- PunchCut out/PunchCut.mp4 --codec=h264 --muted

Download the complete pinned project · Inputs, phase meanings & full source code

Official source / inspected interval

Introducing ChatGPT Atlas

Watch the official OpenAI upload ↗ · Published 2025-10-21T10:22:07-07:00

Inspected 22.5–24.8 s. Consecutive subset: 22.9–23.2 s, 9 frames. Dates come from the official watch-page publication field.

Browser context with sidebar open at source 22.900000 seconds — 22.900000 s · Browser context with sidebar open

Last wide frame at source 23.000000 seconds — 23.000000 s · Last wide frame

First detail frame: no travel between them at source 23.033000 seconds — 23.033000 s · First detail frame: no travel between them

Same composer remains in the detail at source 23.166000 seconds — 23.166000 s · Same composer remains in the detail

Source: OpenAI. Bounded excerpt and annotated stills for the analysis below; not included in the project download.

What changes across the edit

Shot boundary: Frame 690 at 23.000 s is wide; frame 691 at 23.033 s is the tight sidebar composition.
Framing: The sidebar composer becomes the subject. Part of the property page remains at the left as an orientation landmark.
Layer ownership: Page and sidebar share the crop; there is no separate moving callout or title.
Motion: An abrupt scale/position discontinuity. Neighboring frames do not show intervening magnifications.
Typography: Native UI text enlarges with its surface; no independently animated title.
Action order: Established browser → direct crop → readable composer.
Pacing & readable holds: The detail persists through this 24.8 s window; later interaction lies outside it.

Recreation versus observation

Matches

A single-frame crop of exactly the same original artifact, with no tween.

Differs

Focus point, 1.55× default magnification and output hold are chosen. Source crop is not numerically fitted.

Measured

Cut pair: 23.000 / 23.033 s, frames 690 / 691.

Inferred

Remaining page content supplies orientation.

Chosen defaults

All Remotion phase frames, cubic Bézier (.16,1,.3,1), geometry, colors, font choice and reading holds are chosen defaults. None is a recovered OpenAI specification.

{
  "title": "Make room for the idea.",
  "prompt": "Arrange a small exhibition.",
  "result": "Three studies, ready to compare.",
  "label": "EXHIBITION / 03",
  "accent": "#40685b",
  "action": 30,
  "change": 54,
  "settle": 54,
  "hold": 90,
  "focusX": 0.66,
  "focusY": 0.64,
  "zoom": 1.55
}

Increase hold to extend only the final quiet state. Change the phase frames to change choreography; preserve their ordering. Content limits and the exact meaning of each phase are in the handoff.

Exact original geometry, layer ownership and asset contracts · Schedule against an actual recording