AIsphere releases video generation product PixVerse V2, which can generate video clips up to 8 seconds for a single clip and 40 seconds for multiple clips

Aishi Technologyrecently released its video generation productPixVerse V2It's a program based on theAI VideoA large model of innovative tools designed to help users unleash their creative potential, PixVerse V2 utilizes the Diffusion+Transformer (DiT) infrastructure with several technological innovations that make video generation smoother, more consistent, and more fun.

Key features include:

• Spatio-temporal Attention Mechanism:PixVerse V2 introduces a self-developed spatio-temporal attention mechanism, which improves the ability to perceive space and time, especially when dealing with complex scenes.
• Text comprehension:With the multimodal model, PixVerse V2 is able to align text information and video information more accurately, which enhances the model's comprehension and expression.
• Optimized Model Training:Based on the traditional flow model, PixVerse V2 promotes faster and better convergence of the model through weighted loss, improving the overall training efficiency.
• Video Generation Capability:PixVerse V2 supports the generation of multiple video clips at a time, up to 8 seconds for a single clip and up to 40 seconds for multiple clips, while maintaining consistency between clips.
• User-Friendly Functions:PixVerse V2 supports one-click generation of 1-5 consecutive video content, and maintains the consistency of the subject image, screen style and scene elements between segments. In addition, users can also edit the generated results to flexibly replace and adjust the video content.

The Aishi Technology team plans to make several iterations and upgrades in the next 3 months to provide a better AI video generation experience.The goal of PixVerse V2 is to make AI video creation easier and more efficient, whether it's documenting daily life or telling a video story.