Technique / FocusType

Establish the input before typing

Make the target of an instruction clear before the words arrive.

Run the original

6.2 s · 30 fps · silent

Original neutral assets · 6.2 seconds · phase values are chosen defaults, not source timings.

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

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

Official source / inspected interval

A first look at the Codex app

Watch the official OpenAI upload ↗ · Published 2026-02-02T10:03:33-08:00

Inspected 6.5–10.5 s. Consecutive subset: 7.5–8.5 s, 24 frames. Dates come from the official watch-page publication field.

Workspace area before composer reframe at source 6.507000 seconds — 6.507000 s · Workspace area before composer reframe

Composer now occupies the lower frame at source 7.216000 seconds — 7.216000 s · Composer now occupies the lower frame

Empty input remains available at source 8.300000 seconds — 8.300000 s · Empty input remains available

Request growing; submission not shown at source 10.135000 seconds — 10.135000 s · Request growing; submission not shown

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

What changes across the edit

Shot boundary: This excerpt is a reframe and input beat, not evidence of a submission cut.
Framing: The view moves down from the workspace selector toward the composer. The lower window edge remains an orientation cue.
Layer ownership: Native composer, caret/pointer and window chrome; no editorial keystroke overlay.
Motion: Camera reframing precedes the request growth visible from roughly 9 s. The later text does not fully finish in this window.
Typography: Native sentence grows along the composer baseline. Its typing rate is not recovered.
Action order: Target is located → field remains visible → text begins growing.
Pacing & readable holds: Empty input is available across several samples before typing. This interval does not establish complete-request reading time or submission.

Recreation versus observation

Matches

Pointer aim before text growth, stable field and baseline.

Differs

The original example completes a shorter request and holds it. No submission is animated or inferred; camera movement is omitted to isolate focus order.

Measured

Inspected 6.500–10.500 s request, consecutive subset 7.500–8.500 s.

Inferred

Target-first sequencing helps causal legibility; source caret and camera timing are not fitted.

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": 24,
  "change": 54,
  "settle": 126,
  "hold": 60,
  "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