PACER+ is a novel framework designed to generate diverse and realistic pedestrian animations in driving scenarios. It addresses the limitations of existing methods by combining trajectory following with fine-grained control over specific body parts, enabling zero-shot recreation of real-world pedestrian animations. The framework uses a joint training scheme that trains a single policy to track specific body parts while following given trajectories. This approach enhances both the diversity and controllability of simulated human motions, making it suitable for various driving scenarios, including synthetic and real-world environments. Key contributions include: 1. **Unified Physics-Based Framework**: PACER+ combines motion tracking and trajectory following tasks, allowing for the generation of diverse pedestrian behaviors. 2. **Zero-Shot Control**: The framework supports fine-grained control over specific body parts while following given trajectories, enabling the creation of realistic and varied animations. 3. **Versatility**: PACER+ can generate animations from various sources, including generative models, pre-captured motions, and videos, in both manual and real-world scenarios. 4. **Real-World Recreation**: The framework can recreate real-world pedestrian animations without re-training or fine-tuning, filling in missing parts automatically. The evaluation demonstrates that PACER+ outperforms state-of-the-art methods in terms of motion quality, diversity, and tracking accuracy, particularly in low-speed scenarios and complex terrains. The framework's effectiveness is further validated through experiments on synthetic and real-world datasets, showing its ability to generate realistic and diverse pedestrian animations.PACER+ is a novel framework designed to generate diverse and realistic pedestrian animations in driving scenarios. It addresses the limitations of existing methods by combining trajectory following with fine-grained control over specific body parts, enabling zero-shot recreation of real-world pedestrian animations. The framework uses a joint training scheme that trains a single policy to track specific body parts while following given trajectories. This approach enhances both the diversity and controllability of simulated human motions, making it suitable for various driving scenarios, including synthetic and real-world environments. Key contributions include: 1. **Unified Physics-Based Framework**: PACER+ combines motion tracking and trajectory following tasks, allowing for the generation of diverse pedestrian behaviors. 2. **Zero-Shot Control**: The framework supports fine-grained control over specific body parts while following given trajectories, enabling the creation of realistic and varied animations. 3. **Versatility**: PACER+ can generate animations from various sources, including generative models, pre-captured motions, and videos, in both manual and real-world scenarios. 4. **Real-World Recreation**: The framework can recreate real-world pedestrian animations without re-training or fine-tuning, filling in missing parts automatically. The evaluation demonstrates that PACER+ outperforms state-of-the-art methods in terms of motion quality, diversity, and tracking accuracy, particularly in low-speed scenarios and complex terrains. The framework's effectiveness is further validated through experiments on synthetic and real-world datasets, showing its ability to generate realistic and diverse pedestrian animations.