Foreign Media: Chinese Researchers Release Near-Instant AI Video Generation New Technology
Chinese researchers have developed a technology called TurboDiffusion, which can increase the speed of AI video generation by 200 times without reducing image quality. On a consumer-level system equipped with an NVIDIA RTX 5090 graphics card, the time to generate a five-second standard-definition video is reduced from three minutes to 1.9 seconds, and a five-second high-definition video is reduced from nearly 80 minutes to 24 seconds.
Currently, OpenAI's Sora text-to-video model also requires several minutes to render short clips. TurboDiffusion demonstrates China's rapid progress in AI video generation technology and is expected to make video generation models no longer a bottleneck in the production process.
Original article: toutiao.com/article/1852672598263900/
Statement: This article represents the views of the author himself.