Kinetic Typography
Design structured typography animations that sync cleanly with voiceover cadences and preserve multilingual script integrity.
1. Typographic Scale Ladders
To maintain legibility across vertical (9:16) and widescreen (16:9) formats, text size is defined as a fraction of total viewport height:
- Display:
0.08to0.12frame height (14–20 characters per line max) - Headline:
0.05to0.07frame height (24–32 characters per line max) - Subhead:
0.035to0.045frame height (36–44 characters per line max) - Caption / Body:
0.024to0.032frame height (40–50 characters per line max) - Badge / Tag:
0.018to0.025frame height (12–18 characters per line max)
2. Animation Primitives
Text behaviors accept modular entrance, emphasis, and exit configurations:
fade_rise: Characters or words elevate 24px upwards while fading in smoothly.roll_in: Vertical 3D drum rotation where words snap onto the baseline like a slot wheel.highlight_wipe: A high-contrast color block sweeps horizontally across selected hero words.pop_in: Fast scale pop with slight elastic rebound on focal keywords.typewriter: Sequential character reveal with optional blinking vertical cursor (|).
See Animation Primitives Reference for full parameter definitions.
3. Speech-Locked Synchronization (syncTo: "speech")
When aligning captions with real voiceover or STT transcripts, avoid arbitrary uniform staggers. Bind word reveal timestamps directly to audio speech boundaries via spokenWords:
- Extract word-level start and end timestamps from speech recognition data.
- Provide
syncTo: "speech"in the typography configuration. - Each word activates at its exact spoken timestamp (
startSecs), switching into full opacity or accent emphasis.
4. Indic Akshara Syllable Integrity
Standard character-splitting algorithms break Indian scripts (Devanagari, Telugu, Tamil, Bengali, Kannada) by splitting vowel matras and conjunct consonants from their base characters.
The typography engine applies linguistic Akshara grapheme cluster rules:
- Consonant + Halant/Virama + Consonant sequences are preserved as single indivisible animated units.
- Vowel signs and anusvaras remain bonded to the preceding consonant.
5. Declarative MotionSpec Example
{
"$schema": "https://vichith.in/schemas/motion-spec.json",
"layer": "headline-callout",
"type": "text",
"text": "Autonomous video editing with agentic intelligence",
"behavior": {
"unit": "word",
"enter": "roll_in",
"emphasis": "highlight_wipe",
"highlightColor": "#38bdf8",
"staggerSecs": 0.07,
"syncTo": "speech",
"spokenWords": [
{ "text": "Autonomous", "startSecs": 0.12, "endSecs": 0.54 },
{ "text": "video", "startSecs": 0.58, "endSecs": 0.88 },
{ "text": "editing", "startSecs": 0.92, "endSecs": 1.34 },
{ "text": "with", "startSecs": 1.38, "endSecs": 1.55 },
{ "text": "agentic", "startSecs": 1.60, "endSecs": 2.05 },
{ "text": "intelligence", "startSecs": 2.10, "endSecs": 2.68 }
]
}
}