An Open Source High Performance Video Generation Model from a Chinese Company

IBL News | New York

The Chinese company Tencent presented a framework for a large video generation model called “HunyuanVideo.”

This Text-to-Video model was released as open source with its code and weights.

HunyuanVideo was trained with over 13 billion parameters, making it the largest open-source model.

“It exhibits performance in video generation that is comparable to, if not superior to, leading closed-source models,” said the authors.

“It outperforms state-of-the-art models, including Runway Gen-3, Luma 1.6, and 3 top performing Chinese video generative models.”