Motion Frames Skill

Produce a single full-bleed motion composition. Inline CSS animations only — the page is the loop. Treat it as a poster frame that an exporter (HyperFrames, Lottie, etc.) can capture into a video.

Workflow

Read the active DESIGN.md (injected above). Motion lives or dies on typography contrast — pick the most expressive serif / display token in the DS for the headline; the body / mono token labels everything else.
Compose the canvas as a 16:9 hero with these layers, back to front:
- Stage — full-bleed <main>. Off-white or DS-canvas background, very subtle dotted grid texture (CSS background, radial-gradient dots at 22–32px intervals).
- Concentric rings — 2–3 SVG circles radiating from a focal point. Ultra-thin strokes (0.5–1px) in DS-foreground at low opacity. These rotate at different speeds (60s, 90s, 180s).
- Focal mark — a wireframe globe, a stylized object, or a typographic monogram drawn as inline SVG. ~28% of the canvas wide.
- Ring labels — short words / phonetic tokens placed around one of the rings (e.g. "Hola · Bonjour · 你好 · नमस्ते"). They co-rotate with the ring, with <text> paths counter-rotated so the words stay upright.
- Headline — bottom-left or center-bottom. Display serif, italic accent on one word. Add a subtle letterSpacing + opacity reveal animation (@keyframes type-in).
- Frame chrome — corner stamps (top-left lab tag, top-right brand or issue number) and a thin baseline rule. Static.
Animate with @keyframes only — no JS:
- rotate-slow, rotate-med, rotate-fast for rings.
- globe-spin for the focal mark.
- pulse for the focal dot, ~2s, easing.
- marquee-fade to reveal headline once on load.
Write a single HTML document:
- <!doctype html> through </html>, CSS inline.
- All motion uses CSS — no scripts, so HyperFrames or any frame-grabber can capture it deterministically.
- data-od-id on stage, focal, ring, headline, chrome.
Self-check:
- The composition still reads as a poster with motion paused at frame 0.
- At least 3 layers move at different speeds (depth comes from delta velocity, not parallax tricks).
- Accent appears once — usually the italic word in the headline.

Output contract

Emit between <artifact> tags:

<artifact identifier="motion-slug" type="text/html" title="Motion — Title">
<!doctype html>
<html>...</html>
</artifact>

One sentence before the artifact, nothing after.

Workflow

Read the active DESIGN.md (injected above). Motion lives or dies on typography contrast — pick the most expressive serif / display token in the DS for the headline; the body / mono token labels everything else.

Compose the canvas as a 16:9 hero with these layers, back to front:

Stage — full-bleed <main>. Off-white or DS-canvas background, very subtle dotted grid texture (CSS background, radial-gradient dots at 22–32px intervals).
Concentric rings — 2–3 SVG circles radiating from a focal point. Ultra-thin strokes (0.5–1px) in DS-foreground at low opacity. These rotate at different speeds (60s, 90s, 180s).
Focal mark — a wireframe globe, a stylized object, or a typographic monogram drawn as inline SVG. ~28% of the canvas wide.
Ring labels — short words / phonetic tokens placed around one of the rings (e.g. "Hola · Bonjour · 你好 · नमस्ते"). They co-rotate with the ring, with <text> paths counter-rotated so the words stay upright.
Headline — bottom-left or center-bottom. Display serif, italic accent on one word. Add a subtle letterSpacing + opacity reveal animation (@keyframes type-in).
Frame chrome — corner stamps (top-left lab tag, top-right brand or issue number) and a thin baseline rule. Static.

Animate with @keyframes only — no JS:

rotate-slow, rotate-med, rotate-fast for rings.
globe-spin for the focal mark.
pulse for the focal dot, ~2s, easing.
marquee-fade to reveal headline once on load.

Write a single HTML document:

<!doctype html> through </html>, CSS inline.
All motion uses CSS — no scripts, so HyperFrames or any frame-grabber can capture it deterministically.
data-od-id on stage, focal, ring, headline, chrome.

Self-check:

The composition still reads as a poster with motion paused at frame 0.
At least 3 layers move at different speeds (depth comes from delta velocity, not parallax tricks).
Accent appears once — usually the italic word in the headline.

motion-frames

让 AI 帮你安装

Motion Frames Skill

Workflow

Output contract

链接

创作者

Motion Frames Skill

Workflow

Output contract