Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

China's generative video race heats up | TechCrunch

Feb 06, 2024 - techcrunch.com
Chinese internet giant Tencent has released a new version of its open-source video generation model, DynamiCrafter, on GitHub. The model uses a diffusion method to convert captions and still images into short videos, with the second generation offering an improved pixel resolution of 640×1024. The technology is unique in its broad applicability of image animation techniques to more general visual content, using the motion prior of text-to-video diffusion models as guidance.

The release is part of a wider trend of Chinese tech firms, including ByteDance, Baidu, and Alibaba, investing in the text- and image-to-video space. These companies have also released their video diffusion models, with Alibaba's VGen also being open-source. This strategy is becoming increasingly popular among Chinese tech firms aiming to engage the global developer community.

Key takeaways:

  • Tencent has unveiled a new version of its open source video generation model, DynamiCrafter, on GitHub, indicating China's growing efforts in the text- and image-to-video space.
  • DynamiCrafter uses the diffusion method to convert captions and still images into short videos, with the second generation producing videos at a pixel resolution of 640×1024.
  • The technology behind DynamiCrafter differs from competitors as it broadens the applicability of image animation techniques to more general visual content, using the image in the generative process as guidance.
  • Other Chinese tech firms including ByteDance, Baidu and Alibaba have also released their video diffusion models, with Alibaba's VGen also being open source, reflecting a trend among Chinese tech firms to reach the global developer community.
View Full Article

Comments (0)

Be the first to comment!