Upload an image and describe the motion. This Space runs an open-source image-to-video model on Hugging Face ZeroGPU when GPU quota is available.