At the 2024 Volcano Engine AI Innovation Tour, Tan Dai, President of Volcano Engine, unveiled the latest Doubao Video Generation Model.
The model incorporates several advanced technologies, including an efficient DIT fusion computing unit that enables deep compression encoding of video and text; it also adopts a new diffusion model training method, ensuring consistency in generating multi-shot videos; furthermore, the model integrates a deeply optimized Transformer structure, significantly enhancing the generalization capability of video generation.
Tan Dai emphasized during the launch event that the Doubao Video Generation Large Model supports consistent multi-shot generation in multiple styles and proportions, suitable for various fields such as e-commerce marketing, animation education, urban tourism, and micro-script production.
