New framework enables real-time, one-shot emotion-controllable portrait animation

Researchers have developed a new framework called Proxy Avatar Meets Low-Rank Caching for real-time, one-shot portrait animation driven by audio and emotion. This method utilizes a Gaussian-based emotion proxy avatar to generate expressive motion, which is then adapted to target portraits by a one-shot retargeting model. To enhance efficiency, the system employs low-rank caching to reuse appearance features and low-rank adapters to model feature variations, enabling faster inference and real-time animation. AI

IMPACT Enables more expressive and efficient AI-driven animation for digital avatars and content creation.

RANK_REASON This is a research paper detailing a new technical framework for portrait animation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• alphaXiv
• arXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• Gaussian-based emotion proxy avatar
• Gotit.pub
• Hugging Face
• Influence Flower
• Litmaps
• Low-Rank Adapters
• low-rank caching
• one-shot retargeting model
• Proxy Avatar Meets Low-Rank Caching
• ScienceCast
• scite Smart Citations

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →