The Importance of Disentanglement: SAGE Outperforms Unified VQ-VAE Baselines in Full-Body Motion

This article presents an ablation study confirming that disentangling motion latents into upper and lower halves significantly enhances 3D avatar reconstruction accuracy