Learn about the different modes of Earth-2 Weather Analytics Blueprint workflows. The desktop workflow is for developing the application locally and can also be used for deployment as desktop applications. The streaming workflow is for deployment in a client server fashion where you can leverage GPUs to render on the server and stream the pixels to the client broswer based application.
The desktop workflow is the primary deployment method for this blueprint to enable developers to locally deploy and customize the application. In this workflow, the underlying microservices and modules such as the data federation mesh (DFM) and FourCastNet NIM are deployed locally using Kubernetes.
The desktop environment features an Omniverse Kit application bundled with extensions, providing users with an interactive globe as an interface for weather analytics and visualization.
When you interact with the Omniverse Kit application, it communicates your requests to the running DFM instance. Upon receiving a request, the DFM instance executes the corresponding data pipeline and saves the resulting textures to a shared cache location. This shared cache enables seamless communication between the data federation mesh and the Omniverse Kit application. The Omniverse Kit application can then access these cached textures and render them directly onto the interactive globe, creating a fluid visualization experience.
The key difference between the desktop and streaming workflows is that the streaming variant deploys the Omniverse application as a containerized service with a webRTC streaming API.
This webRTC stream can then be integrated with a web application to allow for a fully remote experience.
This streaming workflow is achievable with the provided blueprint, however, it requires additional setup and is not fully documented at this time. We include it here as a reference to illustrate how we created the Blueprint experience on the NVIDIA API catalog.
We encourage users that are interested in the streaming capabilities of Omniverse to refer to the WebRTC Browser Client Documentation for more information.
| Previous | Next |
|---|---|
| Readme | Prerequisites |