Vidu is the latest AI video tool from China, and it’s awesome

The AI video generation landscape continues to evolve with Vidu 1.5, a new platform from Chinese company Shengshu that introduces innovative features to compete with established players like Runway and Kling.

Key Innovation: Vidu 1.5 introduces ‘multi-entity consistency,’ a groundbreaking feature that enables users to combine unrelated images into cohesive videos.

• The technology allows creators to merge separate elements, such as placing a person from one photo into a car from another photo to create fluid video content
• Users can incorporate different clothing or objects from separate images into a single character’s video sequence
• The feature has demonstrated practical applications, such as improving skateboarding motion by incorporating multiple perspective images

Technical Capabilities: Vidu 1.5 offers enhanced control and quality options that set it apart from existing platforms.

• The platform supports 1080p video generation, surpassing the typical 720p limitation of competitors
• Creators can customize motion degree, resolution, duration, and other parameters
• An ‘Advanced Character Control’ feature provides precise management of camera movements and cinematic techniques
• Motion speed can be adjusted between auto, low, medium, and high settings for more authentic outputs

Competitive Position: While showing promise in certain areas, Vidu 1.5 still faces some challenges in the broader AI video market.

• The platform’s text-to-video capabilities currently lag behind competitors like Runway, Kling, and MiniMax
• Traditional text-to-video and image-to-video modes are available, offering both photorealistic and illustration options
• The platform is positioning itself as a challenger to OpenAI’s upcoming Sora release

Industry Perspective: Leadership at Shengshu Technology sees AI video generation as a democratizing force in content creation.

• CEO Jiayu Tang emphasizes the technology’s potential to make high-quality content production accessible to everyone
• The platform builds on recent research showing how AI video models interpret physics from images rather than understanding physical principles

Future Implications: The introduction of multi-entity consistency represents a significant advancement that could reshape AI video generation capabilities and influence future development across the industry, potentially spurring other platforms to develop similar features to remain competitive.