# Cut to black and flash a five-word line one word at a time, then land a small product tag

Manual DaVinci Resolve / Fusion construction guide generated from the recipe. It is not a tested .drp.
The accompanying Remotion project (recipe.json) is the executable definition; the numbers below are copied from it.
Neither construction establishes source fidelity: this is an original vector approximation of an inspected editing moment.

## Source and clock
Reference: OpenAI's "OpenAI Super Bowl 2026 | Codex | You Can Just Build Things" (aCN9iCXNJqQ), published 2026-02-08.
Source interval [1241, 1379) on the 24000/1001 fps master: 138 frames, 5.755750 s. Local frame 0 = source frame 1241.
Create a 1920x1080 progressive timeline at 24000/1001 fps and a Fusion composition exactly 138 frames long. Keep integer frames; do not round the frame rate.
Background field: #050507. Export silent H.264 yuv420p to match the intended playback container.

## Observation
A laptop shot hard-cuts to near-black with a faint starfield (~f2). One bold uppercase word appears centred at ~f3 and holds; it is replaced by the next word at ~f27, then ~f48, ~f72 and ~f99, each a hard swap with no motion. At ~f120 the last word is replaced by a much smaller product name in the same centred position, which holds until it starts erasing just after the window.

## Mechanism
Word-per-beat caption: a single centred slot that hard-swaps through the words of a sentence on irregular beats, then downshifts in scale to a small tag.

## Phases (local frames)
- 0: Laptop shot
- 2: Cut to black
- 3: Word 1
- 27: Word 2
- 48: Word 3
- 72: Word 4
- 99: Word 5
- 120: Small product tag

## Layer inventory (absolute 1920x1080 top-left pixels, back to front)
- desk [rect] at (0, 0) size 1920x1080 fill #2a2218 visible [0, 2)
- laptop [rect] at (520, 220) size 900x560 radius 14 fill #111 visible [0, 2)
- screen [rect] at (560, 250) size 820x500 radius 6 fill #1a1d22 visible [0, 2)
- black [rect] at (0, 0) size 1920x1080 fill #050507 visible [2, 138)
- nebula [ellipse] at (300, 300) size 1400x500 fill #1c1a26 visible [2, 138)
- star-a [ellipse] at (400, 200) size 4x4 fill #ffffff visible [2, 138)
- star-b [ellipse] at (1500, 760) size 3x3 fill #ffffff visible [2, 138)
- star-c [ellipse] at (1200, 180) size 3x3 fill #ffffff visible [2, 138)
- star-d [ellipse] at (700, 880) size 4x4 fill #ffffff visible [2, 138)
- w1 [text] at (0, 460) size 1920x0 text "YOU" 150px weight 700 #ffffff visible [3, 27)
- w2 [text] at (0, 460) size 1920x0 text "CAN" 150px weight 700 #ffffff visible [27, 48)
- w3 [text] at (0, 460) size 1920x0 text "JUST" 150px weight 700 #ffffff visible [48, 72)
- w4 [text] at (0, 460) size 1920x0 text "MAKE" 150px weight 700 #ffffff visible [72, 99)
- w5 [text] at (0, 460) size 1920x0 text "THINGS" 150px weight 700 #ffffff visible [99, 120)
- tag [text] at (0, 516) size 1920x0 text "Forge" 34px weight 600 #ffffff visible [120, 138)

## Construction steps
1. Timeline 1920x1080 at 23.976 fps; 138-frame sequence. Local 0 = master frame 1241.
2. Cut from the live-action shot to a near-black plate (#050507 with a very soft nebula and a few star dots) at frame 2.
3. One Text+ node per word, 150px bold uppercase white, centred with the cap height on the frame centre (top y=460). Clip ranges: 3–26, 27–47, 48–71, 72–98, 99–119.
4. Tag: Text+ 34px semibold centred at top y=516, starting at frame 120 and running past the end.
5. No transitions between words; align the swaps to the music beats when re-timing to a different track.
6. Easy mistake: equal spacing. The beats are 24, 21, 24, 27 and 21 frames, not a metronome.

## Content variants
Default copy:
- w1: YOU
- w2: CAN
- w3: JUST
- w4: MAKE
- w5: THINGS
- tag: Forge
Adaptation changes:
- w1: GO
- w2: AND
- w3: SHIP
- w4: REAL
- w5: WORK
- tag: Anvil

## Limits
Easing names map to Fusion spline shapes: linear = linear, in/out/inOut = ease curves on the same frames, hold = step, spring/overshoot = slight overshoot past the target before settling.
Text reveal frames give the first and last unit; distribute characters/words evenly between them. Cursor click rings last 14 frames.
Vector illustration is an approximation of layout and motion, not the product UI or recovered production assets.