# Hold a caption over live footage, hard-cut to a sticker word, then stack tilted photo cards behind it

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 "Turn the world into cheese (or anything really) with this camera." (09-krjdFfis), published 2026-06-02.
Source interval [1130, 1200) on the 30/1 fps master: 70 frames, 2.333333 s. Local frame 0 = source frame 1130.
Create a 1920x1080 progressive timeline at 30/1 fps and a Fusion composition exactly 70 frames long. Keep integer frames; do not round the frame rate.
Background field: #7fb2e5. Export silent H.264 yuv420p to match the intended playback container.

## Observation
A small white caption 'Turn the world into' sits top-centre over a handheld shot of a parking lot with a lime camera in the foreground (f0–19). At f20 the frame hard-cuts: a large photo card tilted about -6° fills most of the frame with a red, hand-lettered sticker word across it. At ~f33 a second tilted card pops in top-right, ~f38 a third bottom-left and ~f45 a fourth bottom-right, each scaling up quickly; the sticker word stays on top throughout and everything holds to f69.

## Mechanism
A hard cut from live-action caption to a graphic collage: a sticker word sits on a tilted hero card and further rotated cards pop in behind it in a quick stagger.

## Phases (local frames)
- 0: Caption holds over footage
- 20: Hard cut: hero card and sticker word
- 33: Second card pops top-right
- 38: Third card pops bottom-left
- 45: Fourth card pops bottom-right
- 69: Hold

## Layer inventory (absolute 1920x1080 top-left pixels, back to front)
- footage [group] at (0, 0) size 1920x1080 visible [0, 20)
  - sky [rect] at (0, 0) size 1920x540 fill #7fb2e5
  - cloud-1 [rect] at (200, 80) size 420x90 radius 45 fill #e8f1fb
  - cloud-2 [rect] at (1200, 140) size 360x70 radius 35 fill #e8f1fb
  - tower [rect] at (380, 240) size 220x300 radius 6 fill #3f5b5c
  - lot [rect] at (0, 540) size 1920x540 fill #5d6066
  - stripe-0 [rect] at (200, 760) size 160x6 radius 3 fill #d9d9d9
  - stripe-1 [rect] at (500, 760) size 160x6 radius 3 fill #d9d9d9
  - stripe-2 [rect] at (800, 760) size 160x6 radius 3 fill #d9d9d9
  - stripe-3 [rect] at (1100, 760) size 160x6 radius 3 fill #d9d9d9
  - stripe-4 [rect] at (1400, 760) size 160x6 radius 3 fill #d9d9d9
  - stripe-5 [rect] at (1700, 760) size 160x6 radius 3 fill #d9d9d9
  - car [rect] at (900, 600) size 280x90 radius 20 fill #1f2328
  - camera [rect] at (160, 620) size 540x460 radius 30 fill #b9c33a
  - cam-screen [rect] at (320, 760) size 220x150 radius 10 fill #1f1f1f
  - caption [text] at (0, 60) size 1920x0 text "Turn the world into" 34px weight 600 #ffffff
- collage [group] at (0, 0) size 1920x1080 visible [20, 70)
  - collage-bg [rect] at (0, 0) size 1920x1080 fill #6f9fd2
  - card-1 [group] at (330, 200) size 1260x700 visible [20, 70)
      keyframes scale: f20=0.7, f25=1 (out)
      keyframes opacity: f20=0, f22=1 (out)
    - card-1-bg [rect] at (330, 200) size 1260x700 radius 4 fill #79a9dc
    - c1-lot [rect] at (330, 580) size 1260x320 fill #e8c14a
    - c1-tower [rect] at (450, 320) size 180x300 radius 4 fill #f0c85a
    - c1-tower2 [rect] at (670, 400) size 120x220 radius 4 fill #f4d271
    - c1-car [rect] at (950, 670) size 320x110 radius 30 fill #f5c94a
  - card-2 [group] at (1180, -40) size 760x600 visible [33, 70)
      keyframes scale: f33=0.7, f38=1 (out)
      keyframes opacity: f33=0, f35=1 (out)
    - card-2-bg [rect] at (1180, -40) size 760x600 radius 4 fill #5b8fd6
    - c2-bldg [rect] at (1300, 20) size 520x540 radius 6 fill #f2c65a
    - c2-win-0 [rect] at (1340, 70) size 60x60 radius 4 fill #d9a834
    - c2-win-1 [rect] at (1460, 70) size 60x60 radius 4 fill #d9a834
    - c2-win-2 [rect] at (1580, 70) size 60x60 radius 4 fill #d9a834
    - c2-win-3 [rect] at (1700, 70) size 60x60 radius 4 fill #d9a834
    - c2-win-4 [rect] at (1340, 170) size 60x60 radius 4 fill #d9a834
    - c2-win-5 [rect] at (1460, 170) size 60x60 radius 4 fill #d9a834
    - c2-win-6 [rect] at (1580, 170) size 60x60 radius 4 fill #d9a834
    - c2-win-7 [rect] at (1700, 170) size 60x60 radius 4 fill #d9a834
    - c2-win-8 [rect] at (1340, 270) size 60x60 radius 4 fill #d9a834
    - c2-win-9 [rect] at (1460, 270) size 60x60 radius 4 fill #d9a834
    - c2-win-10 [rect] at (1580, 270) size 60x60 radius 4 fill #d9a834
    - c2-win-11 [rect] at (1700, 270) size 60x60 radius 4 fill #d9a834
  - card-3 [group] at (-80, 600) size 700x520 visible [38, 70)
      keyframes scale: f38=0.7, f43=1 (out)
      keyframes opacity: f38=0, f40=1 (out)
    - card-3-bg [rect] at (-80, 600) size 700x520 radius 4 fill #6fa3dd
    - c3-ground [rect] at (-80, 940) size 700x180 fill #e2b84a
    - c3-tower [rect] at (200, 720) size 90x230 radius 4 fill #f4d271
  - card-4 [group] at (1160, 660) size 760x480 visible [45, 70)
      keyframes scale: f45=0.7, f50=1 (out)
      keyframes opacity: f45=0, f47=1 (out)
    - card-4-bg [rect] at (1160, 660) size 760x480 radius 4 fill #5f93d4
    - c4-figure [rect] at (1540, 760) size 140x380 radius 60 fill #f5cb5a
    - c4-figure-head [rect] at (1560, 700) size 100x100 radius 50 fill #f5cb5a
  - sticker [group] at (470, 400) size 980x250
    - sticker-bg [rect] at (470, 400) size 980x250 radius 125 fill #ff4a2a
    - word [text] at (470, 410) size 980x0 text "cheese" 190px weight 800 #111111

## Construction steps
1. Timeline 1920x1080 at 30 fps; Fusion comp 70 frames. Local 0 = master frame 1130.
2. Frames 0–19: the live plate (here blocked in with sky, lot, car and lime camera shapes) with a 34px white caption centred at y 60. No animation.
3. Frame 20: hard cut. Enable the collage group and the sticker; disable the plate on the same frame.
4. Hero card: 1260x700 rotated -6° with a drop shadow, Size 0.7 to 1.0 ease-out over frames 20–25. Cards two to four (760x600 at +12°, 700x520 at -14°, 760x480 at +8°) use the same 5-frame pop starting at 33, 38 and 45.
5. Sticker: a 980x250 #ff4a2a pill rotated -4° holding the word at 190px weight 800 with -6 tracking; keep it above every card.
6. Easy mistake: dissolving into the collage; the source is a hard cut and the cards pop, they do not fade.

## Content variants
Default copy:
- caption: Turn the world into
- word: cheese
Adaptation changes:
- caption: Turn your street into
- word: candy

## 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.