AI Video #news

ByteDance launches Vidi, an AI model that can analyze, understand, and edit ultra-long videos. A new era of intelligent video editing, with advanced tools for creators, educators, and filmmakers.

MAGI-1 is an open-source AI model developed by SandAI to generate realistic, long and controllable videos. Based on an autoregressive Diffusion Transformer architecture, it offers top-level performance and is ideal for creative and professional applications.

FramePack is a new AI architecture that allows you to generate quality video directly locally with only 6GB of VRAM. A revolutionary solution for those who want to access AI video without expensive hardware.

UniAnimate-DiT is a powerful open-source tool for transforming static images into smooth and coherent video animations, thanks to an advanced diffusion transformer model. A perfect ally for content creators, developers and digital artists.

Seaweed-7B is the new video model from ByteDance that challenges giants like OpenAI and Google with only 7 billion parameters. Compact but powerful, it is perfect for creative and commercial applications thanks to its efficiency, realism and versatility.

Discover HumanRig, Amap's innovative technology for automatic 3D character rigging, along with the industry's first large open-source dataset. A technological leap that promises to revolutionize animation, accelerate production, and give more creative power to users.

With Gemini 2.5 Pro, Google offers a complete AI workflow for video creation. From content generation to music and editing, everything is integrated and automated for professional results.