Request: Multi-view Cosmos 3 Generator for automotive/robotics via CLI or Python scripts (non-vLLM)

[reikote]

Hi Cosmos team,

I want to use the Cosmos 3 Generator for multi-view video generation in automotive use cases (and potentially robotics), but my assumption is to run it through CLI or Python scripts, similar to this repository, not via vLLM serving.

I found a related vLLM-based example in another repository

However, what I am specifically looking for is guidance for non-vLLM workflows:

1. Recommended CLI or Python-script entry points for multi-view generation
2. Required input format/configuration for automotive or robotics multi-view scenarios
3. GPU requirements for this workflow (minimum and recommended)

Do you have plans to publish code or documentation in this repository for automotive/robotics-oriented multi-view generation using CLI/Python scripts?

If this is already on the roadmap, a rough timeline or pointer to upcoming docs/examples would be very helpful.

Thank you.