Alibaba Cloud recently unveiled the AI model suite for creating videos, Wan 2.1. This innovative tool allows users to produce various video formats including text-to-video, image-to-video, video editing, text-to-image, and audio extraction from videos. The smallest model in this lineup is the T2V-1.3B, capable of running on NVIDIA 4090 cards and delivering video resolutions up to 720P. However, the main model, sized at 14B, requires larger cards like H100/H800 or multiple cards working in conjunction.
The team evaluated video performance against rival AI models, achieving victory against 3 out of 4 models. Wan’s standout features include seamless video transitions, realistic motion physics, and overall testing scores surpassing OpenAI’s Sora.
To access all models, they can be downloaded from HuggingFace or view videos created using Wan on VividHubs.
TLDR: Alibaba Cloud introduces Wan 2.1 AI model suite for versatile video creation, surpassing competitors in video performance evaluation. Access models on HuggingFace or view Wan-created videos on VividHubs.
Leave a Comment