Real-Time Images, Asynchronous Video: One Container, Two Endpoints

Generating an image and then animating it into a short video can be done on Amazon SageMaker AI, and the deployment pattern in AWS's walkthrough splits the job in two. Two endpoints are deployed from the same AWS vLLM-Omni Deep Learning Container: a real-time endpoint for image generation and an asynchronous endpoint for video generation. The split follows the workloads — the image model returns its result inline, while the longer-running video model writes its output instead of handing it straight back. That is a reasonable shape for any pipeline where one step finishes in seconds and the next takes long enough that you would rather not hold a connection open for it.