Modal Kinetic Typography: FE Vibration Modes + Frozen Video Diffusion for Smooth Glyph Animation
Maham Tanveer · hf · 2026-10-05
Researchers present modal kinetic typography, animating vector glyphs to express semantic concepts while staying legible, by building motion from the glyph's natural vibration modes.
- A finite-element eigenproblem assembled from the vector outline yields the glyph's softest modes (whole letter and per part), with zero-energy rigid translations/rotations applied in closed form;
- A frozen video diffusion model supervises only the modes' amplitudes and phases.
Compared to free-form SDS point optimization, modes are smooth along the outline and driven by few whole-cycle harmonics, avoiding outline tearing and jitter; unlike skeleton/keypoint priors, modes come from the glyph itself (the only prior is an LLM-generated part list). Shape and motion are disentangled by construction, enabling frozen-shape animation. Human raters preferred it over Dynamic Typography, AniClipart, and Astra (GPT-6).
More from Multimodal
- AI-generated parody videos are now nearly indistinguishable from real ones — minchoi · 2026-10-06
- Geometry Duo for ComfyUI brings flicker-free depth and normal guidance to video diffusion — Narrow-Particular202 · 2026-10-06
- Free live TTS: an Edge TTS script that streams sentence-by-sentence with parallel generation — Speedk4011 · 2026-10-06
- Running MiniMax H3 locally on a 12GB GPU: 0.6MP at 4:3 is the sweet spot — vortis23 · 2026-10-06
- Opus 5.5 writes and directs a short film about addiction, rendered in Python — danjr000 · 2026-10-05
- Speakrail: a fully-local open-source full-duplex voice assistant running on a single RTX 4090 — danil_rootint · 2026-10-05