Nvidia launched VideoLDM, which can generate 4.7-second video based on text.

CTOnews.com, April 20, Nvidia and Cornell University's research team recently launched a model called VideoLDM, which can automatically generate videos with the highest resolution of 2048mm, 1280s, 24 frames and 4.7s based on text descriptions.

Nvidia says the model has 4.1 billion parameters, 2.7 billion of which are video-trained, which meets the standards of modern generative AI. CTOnews.com learned from the blog that Nvidia said that through an efficient potential diffusion model (LDM), it is possible to create diversified, high-quality, high-definition videos.

The model can also create a video of a driving scene with a resolution of 1024 x 512 pixels and a maximum of 5 minutes. Nvidia said the project is currently in the research stage and will not be open to the public for the time being.

Detailed reports can be accessed at: https://research.nvidia.com/labs/toronto-ai/VideoLDM/