This PR adds a comprehensive model card for the LongVie 2 model, improving its discoverability and documentation on the Hugging Face Hub.
The updates include:
- Addition of `pipeline_tag: text-to-video` and `license: unknown` metadata.
- A concise description of the model.
- Links to the official paper: [LongVie 2: Multimodal Controllable Ultra-Long Video World Model](https://huggingface.co/papers/2512.13604).
- A link to the project page: https://vchitect.github.io/LongVie2-project/.
- A link to the GitHub repository for code: https://github.com/Vchitect/LongVie.
- Inclusion of a link to the Hugging Face demo space and a GIF for visual context.
- A "Quick Start" section with installation, weight download instructions, and an inference command, directly sourced from the GitHub README.
- The relevant BibTeX citation.
Please review and merge this PR.
