S$^2$AG-Vid: Enhancing Multi-Motion Alignment in Video Diffusion Models via Spatial and Syntactic Attention-Based Guidance

Open in new window