# Type a lower third in two columns: name at left, role and affiliation stacked at right

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 "Meet a mathematician solving previously unsolvable math problems with GPT-5.6" (5X5CALvYvp0), published 2026-07-09.
Source interval [527, 613) on the 24000/1001 fps master: 86 frames, 3.586917 s. Local frame 0 = source frame 527.
Create a 1920x1080 progressive timeline at 24000/1001 fps and a Fusion composition exactly 86 frames long. Keep integer frames; do not round the frame rate.
Background field: #9a9384. Export silent H.264 yuv420p to match the intended playback container.

## Observation
Three frames of a warm close-up, then a hard cut (~f3) to a wide interview shot in front of bookshelves. From about f4 the interviewee's name types on at the lower-left in small bold white type; from f6 a second column starts typing the role on the same baseline about 230px to the right, and from f10 a second, right-aligned affiliation line types beneath the role. All text is complete by f22 and holds until a cut at f86.

## Mechanism
A typewriter lower third laid out as two columns: the name column and a two-line role column type on with overlapping starts, so the block assembles left to right.

## Phases (local frames)
- 0: Close-up tail
- 3: Hard cut to interview
- 4: Name types on (left column)
- 6: Role types on (right column, line 1)
- 10: Affiliation types on (right column, line 2)
- 22: Hold

## Layer inventory (absolute 1920x1080 top-left pixels, back to front)
- closeup [group] at (0, 0) size 1920x1080 visible [0, 3)
  - cu-bg [rect] at (0, 0) size 1920x1080 fill #7a6a5c
  - cu-shape [ellipse] at (-200, -100) size 1400x1300 fill #b8987e
- wide [group] at (0, 0) size 1920x1080 visible [3, 86)
  - shelves [rect] at (0, 0) size 1920x1080 fill #9a9384
  - shelf1 [rect] at (0, 120) size 1920x14 fill #e4dfd3
  - shelf2 [rect] at (0, 420) size 1920x14 fill #e4dfd3
  - shelf3 [rect] at (0, 720) size 1920x14 fill #e4dfd3
  - books1 [rect] at (80, 140) size 700x280 fill #a7a598
  - books2 [rect] at (1100, 140) size 740x280 fill #b0aa9a
  - books3 [rect] at (80, 440) size 500x280 fill #a3a496
  - books4 [rect] at (1300, 440) size 540x280 fill #aaa79b
  - chair [rect] at (760, 360) size 400x720 radius 80 fill #2d2f2e
  - subject [ellipse] at (700, 320) size 520x1000 fill #5a1f2a
- name [text] at (60, 950) text "Marek Zieliński" 22px weight 700 #ffffff reveal chars frames 4-12
- role [text] at (290, 950) text "Computational Mathematician" 22px weight 600 #ffffff reveal chars frames 6-16
- affil [text] at (290, 980) size 700x0 text "Institute of Mathematics / Kraków Research Centre" 22px weight 500 #ffffff reveal chars frames 10-22

## Construction steps
1. Timeline 1920x1080 at 23.976 fps; Fusion composition 86 frames. Local 0 = master frame 527; the cut to the wide shot is at local 3.
2. Three Text+ nodes, all 22px white: name (weight 700) left-anchored at (60,950); role (weight 600) left-anchored at (290,950); affiliation (weight 500) right-justified to x=990 at y=980 so its right edge aligns with the end of the role line.
3. Write On in character mode: name 0 to 1 over frames 4-12; role over 6-16; affiliation over 10-22. No caret.
4. Check the affiliation width against the role: the source lets the second line be wider than the first and right-aligns it, so the block reads as a ragged-left column.
5. Easy mistake: typing the three fields strictly one after another. The starts overlap by two to four frames so the block assembles as one gesture.
6. Hold to frame 86; the interval ends on a hard cut to b-roll.

## Content variants
Default copy:
- name: Marek Zieliński
- role: Computational Mathematician
- affil: Institute of Mathematics / Kraków Research Centre
Adaptation changes:
- name: Ana Ferreira
- role: Number Theorist
- affil: Lisbon Institute of Pure Mathematics

## 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.