SC4D is a video-to-4D generation framework that decouples motion and appearance to achieve superior video-to-4D generation. The framework uses sparse control points and dense 3D Gaussians to generate dynamic 3D objects with high quality and efficiency. To address shape degeneration issues, SC4D introduces Adaptive Gaussian (AG) initialization and Gaussian Alignment (GA) loss. These techniques ensure the fidelity of the learned motion and shape. Additionally, SC4D enables motion transfer onto diverse 4D entities based on textual descriptions. The framework is evaluated on a dataset of synthetic and real-world videos, and it outperforms existing methods in terms of reference view alignment, spatio-temporal consistency, and motion fidelity. The results show that SC4D achieves high-quality 4D generation and effective motion transfer. The method is efficient and can be applied to various domains such as AR/VR, filming, animation, and simulation. The framework is implemented using an MLP and a set of sparse control points, and it is trained on a single Tesla V100 GPU. The results demonstrate that SC4D is a promising approach for video-to-4D generation and motion transfer.SC4D is a video-to-4D generation framework that decouples motion and appearance to achieve superior video-to-4D generation. The framework uses sparse control points and dense 3D Gaussians to generate dynamic 3D objects with high quality and efficiency. To address shape degeneration issues, SC4D introduces Adaptive Gaussian (AG) initialization and Gaussian Alignment (GA) loss. These techniques ensure the fidelity of the learned motion and shape. Additionally, SC4D enables motion transfer onto diverse 4D entities based on textual descriptions. The framework is evaluated on a dataset of synthetic and real-world videos, and it outperforms existing methods in terms of reference view alignment, spatio-temporal consistency, and motion fidelity. The results show that SC4D achieves high-quality 4D generation and effective motion transfer. The method is efficient and can be applied to various domains such as AR/VR, filming, animation, and simulation. The framework is implemented using an MLP and a set of sparse control points, and it is trained on a single Tesla V100 GPU. The results demonstrate that SC4D is a promising approach for video-to-4D generation and motion transfer.