# Hard-swap a subtitle line so its black plate re-sizes to the new text, then clear it on the picture cut

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 "Inside Cricket’s Smartest Backroom | Rajasthan Royals | ChatGPT @rajasthanroyals" (0XPk_MAwCW4), published 2026-08-13.
Source interval [3725, 3794) on the 25/1 fps master: 69 frames, 2.760000 s. Local frame 0 = source frame 3725.
Create a 1920x1080 progressive timeline at 25/1 fps and a Fusion composition exactly 69 frames long. Keep integer frames; do not round the frame rate.
Background field: #241a14. Export silent H.264 yuv420p to match the intended playback container.

## Observation
A seated interviewee in a dark wood-panelled room speaks to camera; at the bottom centre a short subtitle sits on a tight black plate. Frames 0-25 hold that short line. At local 26 (master 3751) the line swaps in one frame to a much longer sentence and the black plate is correspondingly wider, still centred at the same baseline. The long line holds until local 67 (3792), where a hard cut to a tight close-up of the same person arrives with no subtitle at all for the last two frames.

## Mechanism
Subtitle hard swap with a fitted plate: the plate width follows the text on the swap frame, and the subtitle is cut out with the shot rather than carried over.

## Phases (local frames)
- 0: Short subtitle on a narrow plate
- 26: Hard swap to long line; plate widens to fit
- 67: Hard cut to close-up; subtitle cleared

## Layer inventory (absolute 1920x1080 top-left pixels, back to front)
- shot-a [group] at (0, 0) size 1920x1080 visible [0, 67)
  - panel-l [rect] at (560, 0) size 300x1080 fill #5a3b26
  - gap [rect] at (860, 0) size 200x1080 fill #c9b59a
  - panel-r [rect] at (1060, 0) size 300x1080 fill #5a3b26
  - frame-r [rect] at (1420, 80) size 380x300 fill #161210
  - body [rect] at (540, 560) size 840x760 radius 250 fill #1f2a5a
  - stripe [rect] at (690, 620) size 540x44 radius 22 fill #b3223a
  - head [ellipse] at (800, 180) size 320x380 fill #b98462
  - hair [rect] at (790, 150) size 340x130 radius 65 fill #1a1412
- plate-a [rect] at (790, 866) size 340x44 radius 4 fill #111111 visible [0, 26)
- cap-a [text] at (790, 873) size 340x0 text "so from there we lean on" 24px weight 400 #ffffff visible [0, 26)
- plate-b [rect] at (710, 866) size 500x44 radius 4 fill #111111 visible [26, 67)
- cap-b [text] at (710, 873) size 500x0 text "to sharpen how we bid on auction day." 24px weight 400 #ffffff visible [26, 67)
- shot-b [group] at (0, 0) size 1920x1080 visible [67, 69)
  - b-wall [rect] at (0, 0) size 1920x1080 fill #3a2a20
  - b-panel [rect] at (0, 0) size 520x1080 fill #c9b59a
  - b-body [rect] at (700, 760) size 1400x700 radius 300 fill #1f2a5a
  - b-head [ellipse] at (900, 60) size 720x840 fill #b98462
  - b-hair [rect] at (880, 0) size 760x260 radius 130 fill #1a1412

## Construction steps
1. Timeline 1920x1080 progressive at 25 fps; Fusion composition 69 frames. Local 0 = master frame 3725.
2. Shot A stand-in (frames 0-66): dark room with two wood panels and a bright gap, a navy jersey block with a red diagonal stripe and a head ellipse. Shot B stand-in from frame 67: a much larger head filling the right two thirds. Straight cut, no transition.
3. Subtitle A: Background 340x44 #111111 at (790,866) with a 24 px white centred Text+ at y=873; frames 0-25 only. Subtitle B: Background 500x44 at (710,866), same text style; frames 26-66 only. Both plates share the same vertical position and centre line x=960.
4. Size each plate to its text plus ~18 px side padding; do not animate the width. The change is a hard swap on frame 26.
5. End both subtitle nodes at frame 66 so the close-up at 67 arrives clean.
6. Easy mistake: keeping one fixed-width plate. The plate must hug each line, so the visible change at 26 is both new words and a wider black bar.

## Content variants
Default copy:
- lineA: so from there we lean on
- lineB: to sharpen how we bid on auction day.
Adaptation changes:
- lineA: and after that we use it
- lineB: to decide which players we chase first.

## 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.
