# Pop the first word of a lower caption, then type the rest character by character

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 "First impressions of GPT-6 Astra from developers" (-TTyyY3VWh8), published 2026-09-03.
Source interval [126, 174) on the 24000/1001 fps master: 48 frames, 2.002000 s. Local frame 0 = source frame 126.
Create a 1920x1080 progressive timeline at 24000/1001 fps and a Fusion composition exactly 48 frames long. Keep integer frames; do not round the frame rate.
Background field: #4b4f52. Export silent H.264 yuv420p to match the intended playback container.

## Observation
Over an interview shot the bottom-left caption is empty for the first six frames (master 126-131). At master 132 (local 6) the word 'with' appears whole in white bold sans. From local 11 the remaining names type in a few characters per frame ('Ben, Pe' at local 11, 'Peter,' at 15, 'and' at 17, 'and T' at 19) and the full line 'with Ben, Peter, and Tom' is complete by local 21, then holds to the end of the window.

## Mechanism
Caption build in two speeds: the lead word pops on as a unit, then the rest of the line types on character by character and holds.

## Phases (local frames)
- 0: Interview plate, no caption
- 6: Lead word pops on whole
- 11: Names type on by character
- 21: Caption complete, hold

## Layer inventory (absolute 1920x1080 top-left pixels, back to front)
- window-light [ellipse] at (900, -200) size 1500x1300 fill #8b9093
- window-mullion-a [rect] at (1180, 0) size 22x1080 fill #5f6467
- window-mullion-b [rect] at (1560, 0) size 22x1080 fill #5f6467
- foliage [ellipse] at (1180, 320) size 520x360 fill #2b4a3a
- subject-shoulder [ellipse] at (120, 560) size 1100x900 fill #3a3d40
- caption-lead [text] at (96, 962) text "with" 46px weight 600 #ffffff visible [6, 10)
- caption-full [text] at (96, 962) text "with Mara, Devon, and Kai" 46px weight 600 #ffffff reveal chars frames 8-21 visible [10, 48)

## Construction steps
1. Timeline 1920x1080 at 23.976 fps; Fusion composition 48 frames. Local 0 = master frame 126.
2. Background: solid #4b4f52 with a large soft ellipse (#8b9093, 55%) upper right and two vertical mullion bars to read as a bright window; a dark ellipse lower left stands in for the interview subject. No photo.
3. Caption A: Text+ 'with', 46 px, weight 600, white, positioned x 96 / y 962 (top-left of text box). Visible frames 6-9 only, no animation: it pops on.
4. Caption B: Text+ full line at the same position. Write On keyframed 0 at frame 8 to 1 at frame 21 (linear) so the first four characters are already on when it takes over from Caption A at frame 10.
5. Hold the completed caption from frame 21 to 48; nothing else moves.
6. Easy mistake: do not ease the write-on or start it at frame 10; the lead word must never flicker between the two text nodes, so B's write-on has to begin two frames before A hides.

## Content variants
Default copy:
- lead: with
- caption: with Mara, Devon, and Kai
Adaptation changes:
- lead: alongside
- caption: alongside Priya, Noor, and Sam

## 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.