Conceitos essenciais
GAIA aims to eliminate domain priors in talking avatar generation, achieving superior naturalness, diversity, and scalability.
Estatísticas
前のモデルよりも優れた結果を達成します。
フレームごとに潜在表現をエンコードします。
大規模なデータセットから多様性を確保します。
Citações
"GAIA reveals two key insights: 1) the speech only drives the motion of the avatar; 2) there exists enormous diversities in expressions and head poses."
"We propose to eliminate these heuristics and generate the full motion latent at the same time."