# Land a top and bottom title in two beats, then hold it while the footage hard-cuts every 13 frames

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 "How Enterprise Teams Use ChatGPT Work | Zapier, Shopify & Virgin Atlantic" (zq88iLsSfMA), published 2026-07-23.
Source interval [138, 246) on the 24/1 fps master: 108 frames, 4.500000 s. Local frame 0 = source frame 138.
Create a 1920x1080 progressive timeline at 24/1 fps and a Fusion composition exactly 108 frames long. Keep integer frames; do not round the frame rate.
Background field: #8a8f6a. Export silent H.264 yuv420p to match the intended playback container.

## Observation
A webcam-style speaker shot holds for two frames; a large white first title line pops on across the top of frame (~f2), then a second line pops on across the bottom (~f5). The title then stays fixed while the footage hard-cuts through eight different speakers at roughly 13–14 frame intervals (local ~f11, 25, 39, 52, 66, 79, 93, 107). The type never moves or fades; only the plate beneath it changes.

## Mechanism
A locked two-line split title that persists as a constant while the underlying footage jump-cuts on a steady beat.

## Phases (local frames)
- 0: Speaker A, no title
- 2: Top line pops on
- 5: Bottom line pops on
- 11: Cut to speaker B; title holds
- 25: Cut C
- 39: Cut D
- 52: Cut E
- 66: Cut F
- 79: Cut G
- 93: Cut H
- 107: Cut back to A

## Layer inventory (absolute 1920x1080 top-left pixels, back to front)
- plate-0 [rect] at (0, 0) size 1920x1080 fill #8a8f6a visible [0, 11)
- figure-0 [ellipse] at (560, 260) size 800x1100 fill #2b2723 visible [0, 11)
- plate-1 [rect] at (0, 0) size 1920x1080 fill #4b5563 visible [11, 25)
- figure-1 [ellipse] at (697, 260) size 800x1100 fill #1f1f22 visible [11, 25)
- plate-2 [rect] at (0, 0) size 1920x1080 fill #7c8a8f visible [25, 39)
- figure-2 [ellipse] at (594, 260) size 800x1100 fill #2b2723 visible [25, 39)
- plate-3 [rect] at (0, 0) size 1920x1080 fill #3b3b40 visible [39, 52)
- figure-3 [ellipse] at (731, 260) size 800x1100 fill #1f1f22 visible [39, 52)
- plate-4 [rect] at (0, 0) size 1920x1080 fill #6b6f75 visible [52, 66)
- figure-4 [ellipse] at (628, 260) size 800x1100 fill #2b2723 visible [52, 66)
- plate-5 [rect] at (0, 0) size 1920x1080 fill #9a3d3d visible [66, 79)
- figure-5 [ellipse] at (765, 260) size 800x1100 fill #1f1f22 visible [66, 79)
- plate-6 [rect] at (0, 0) size 1920x1080 fill #c9cbc7 visible [79, 93)
- figure-6 [ellipse] at (662, 260) size 800x1100 fill #2b2723 visible [79, 93)
- plate-7 [rect] at (0, 0) size 1920x1080 fill #d7dcd9 visible [93, 107)
- figure-7 [ellipse] at (799, 260) size 800x1100 fill #1f1f22 visible [93, 107)
- top [text] at (0, 36) size 1920x0 text "Teams rely on" 104px weight 600 #ffffff visible [2, 108)
- bottom [text] at (0, 900) size 1920x0 text "Workbench Pro" 104px weight 600 #ffffff visible [5, 108)

## Construction steps
1. Timeline 1920x1080 at 24 fps; 108-frame sequence. Local 0 = master frame 138.
2. Video track 1: eight speaker clips cut end to end at frames 0, 11, 25, 39, 52, 66, 79, 93 and 107 (13–14 frames each); straight cuts, no transitions.
3. Video track 2: one Text+ title clip spanning the whole sequence. Top line 104px semibold white centred, top y=36; bottom line same style, top y=900.
4. Trim the top-line layer to start at frame 2 and the bottom-line layer at frame 5 (two separate Text+ nodes or two clips), no fades.
5. Because the title lives on its own track above the cuts it does not need re-keying per shot; check it reads over both the lightest and darkest plates (add a 40% drop shadow if not).
6. Easy mistake: putting the title inside each clip. It must be a single continuous layer so the cuts happen under it.

## Content variants
Default copy:
- top: Teams rely on
- bottom: Workbench Pro
Adaptation changes:
- top: Crews trust
- bottom: Fieldkit Cloud

## 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.
