It’s a day that ends in Y, so a new AI video generator is joining the growing throng of similar tools. Alibaba is the latest to join the field with its new text-to-video model, part of its Tongyi Wanxiang portfolio. The AI video tool, announced at the Alibaba Cloud Apsara Conference, was just one part of an avalanche of new AI options from the Chinese tech giant, including more than 100 new large language models (LLMs),
Tongyi Wanxiang is Alibaba’s suite of synthetic media generation models, which started last year with an AI image maker. The new tool will produce high-quality videos from text prompts in both Chinese and English, as well as still images. Alibaba executives boasted that the company has one of the most advanced Diffusion Transformer (DiT) architectures, allowing it to create videos that maintain their quality regardless of the style requested by the user, including realistic live-action and many animation styles.
Alibaba didn’t spend too much time on how it imagines users using the AI video maker, but the company’s emphasis on third-party partnerships is suggestive. The technology could be used in a range of marketing and entertainment videos. It could also end up in video games, where visual references or even entire introductory videos are produced.
Have you seen Sora?
The number of AI video generators coming out or coming out soon is staggering, considering there weren’t any consumer-level ones not too long ago. OpenAI garnered a lot of attention for the idea with its Sora model. Still, the company’s decision to limit Sora to certain partners left many people scrambling for alternatives, and companies like Alibaba are happy to fill some of the gaps.
Runway, Stability AI, Pika, Hotshot and Luma Labs’ Dream Machine are just a few of the most prominent examples. And Alibaba isn’t the only Chinese competitor. Kling and TikTok owner Bytedance’s Jimeng are also in the race. Alibaba has announced action, but the final version with eventual winners and losers has yet to be filmed.