Picture Lock / 03 · Motion graphics
4 looks,
none of them a template.
Graphics are drawn per video against the words that were actually said — a stat card shows the number in the audio and its label, never a made-up line on a made-up chart.
Each look was built from something, not sketched. Two of them take identical props, so the same clip can be rendered either way without rewriting the plan — which is the only reason a "look" is worth having as a concept at all.
The looks
Dark grid ground, the speaker cut out by a per-frame matte and standing on it. When a graphic lands the speaker shrinks to the bottom-left and the graphic takes the top 60%.
A paper board, a map the camera travels across, black-and-white sticker cut-outs, yellow highlighter, ink captions with a marker on the spoken word.
Textured paper, a real map, cut-in objects — a membership card, a newspaper clipping — marker strokes that draw on, stamps, and captions that spring in word by word.
A person talks to camera and graphics pop up in the explainer's style around them, on the same props as Desk, so the server can pick either look for the same clip.
The rules every one of them obeys
- top of frame
269px - bottom of frame
672px - each side
65px
- one secondary opacity
28% - frame
1080×1920 - frame rate
30fps
The bottom 35% of the frame is kept clear because that is where the apps draw their own furniture. A graphic that looks right in the editor and is covered by a like button in the feed is not a graphic that worked.
Read straight out of the overlay
system by build_graphics_page.py — last read 27 September 2026.