Welcome to the Team

Training

# Canva Sample — Welcome to the Team A four-scene employee-onboarding film recreated from a screen recording of a Canva design: one purple stadium per scene, running off one edge of the frame, with the copy set inside it over full-bleed footage. Three aspect ratios, four colour schemes, and **one motion idea used four ways** — the stadium fades up, then every piece of copy in the scene fades up together. ## Where it came from The .mov is 26.0s at 60fps of a 890×486 player window, and it loops: the four pages play once and the first begins again, so the whole film is in the recording twice over. Cuts are hard — one frame between the last frame of one page and the first of the next — and every page ends by fading its own copy out to leave the bare footage for a beat. **Those exits are not ported** (motion.md rule 5): they are keyed to a page length the narration replaces, so a filled scene would fade out on schedule and then talk over an empty frame. Nothing here is anchored to a scene's end, and the scenes carry no `transition`. | # | scene | source page | measured | | --- | --- | --- | --- | | 1 | Cover — stadium off the left | "Welcome to the team!" | 4.5s | | 2 | Beat — stadium off the right | "Your onboarding kit" | 5.0s | | 3 | Beat — stadium off the left, one paragraph | "Your new employee kit contains…" | 7.0s | | 4 | Close — low stadium bar | "Ask your supervisor…" | 5.0s | The recording is a crop of the 16:9 page — 890×486 is 1.83:1 — so geometry was converted as `x → (x−2)·1920/888` and `y → (y+6.75)·1920/888`, i.e. assuming the ~13px missing from the height was taken evenly off the top and bottom. The corroboration is that all four stadiums then land their bottom edge on 90.1% of the canvas, which is the guide the designer was snapping to. ## The stadium Every scene is the same shape at a different size: a box with `radius: 100` (a stadium, because the height is always the shorter side) whose off-frame end is never seen, so only one round cap is ever drawn. The fill is `#5A21B0CC` — solid purple at 80%, recovered by regressing the pixels just inside each stadium's top edge against the pixels just above it (slope `1−α` ≈ 0.20 on all three channels in all four scenes), which is why the footage stays legible through it. | scene | edge it runs off | height | cap radius | | --- | --- | --- | --- | | welcome | left | 67.7% | 365px | | kit | right | 49.3% | 266px | | essentials | left | 54.5% | 294px | | support | right (a low bar) | 22.4% | 121px | ## Type Both faces were identified by measurement, not by eye — cumulative advances fitted to the word-gap positions along each measured line: - **Raleway 400** for every piece of body copy. On "our office and our company" the fitted word boundaries land within 1.1px of the measured ones (rms 0.56px); the nearest other candidate is 2× worse, and its cap-to-x-height ratio (1.37) matches the measured 1.378. - **Fraunces 500** for the two serif headings. The ink-width-per-cap-height of "Welcome to" (7.96) and "the team!" (6.22) sit within 1% of Fraunces at 500, and it also wins a raster cross-correlation against the frame. Playfair Display — the obvious guess for a Canva serif — is 7% too narrow at every weight. Sizes at 1920×1080: the welcome is 150px on a 160px line, the kit heading 81px, the paragraph and the sign-off 56px, the standfirst 43px, the wordmark 27px with 0.05em of tracking. Each text box carries `height: -1`, so a longer fill grows the box downward from a pinned top rather than re-centring the copy. ## The motion, and what it is not Measured by high-pass-correlating each frame's copy against the same copy at rest, which separates the type's opacity from the footage moving behind a translucent stadium: | | starts | duration | ease | | --- | --- | --- | --- | | the stadium | 0.02s | 0.45s | `power1.out` | | every text layer | 0.70s | 0.20s | `power1.out` | There is no stagger, no slide and no scale anywhere in the film — the stadium's mask grows only because a fading round cap crosses a threshold late, and the copy's ink centroid is stationary to ±0.5px through its whole fade. All four scenes use the same two numbers, which is what makes the recording read as one design rather than four. ## The other ratios 9:16 and 1:1 are re-laid-out, not scaled. The stadium becomes a bottom-anchored band — bottom edge at 90% of the height, as in landscape — and the copy stacks inside it: off the left in the cover and the paragraph, off the right in the kit and the sign-off, so the mirror the landscape pair sets up survives. Type is recomputed against each canvas (`font.size` is resolution-independent, so the portrait canvas renders any stored number 1.78× larger); the wordmark is set at 34px in portrait because 27px is below the floor of the scale on a 1920-tall canvas. ## Filling it Every scene narrates and every scene's photograph is marked `broll`, so a workflow cuts b-roll under the voiceover rather than holding one still for as long as the script runs. The slots are `headline` / `subline` / `detail` / `backdrop`, with `maxLength` set from the box each one was measured in; the wordmark is `ignoreCopy`, because a brand line is typed, not generated.