It’s a day ending in Y, so a new AI video generator is joining the ever-growing mass of similar tools. Alibaba is the latest to join the field with its new text-to-video model, part of its Tongyi Wanxiang portfolio. Announced at the Alibaba Cloud Apsara Conference, the AI video tool was only 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 collection of synthetic media generation models, starting with an AI image creator last year. The new tool will produce high-quality videos from text prompts in both Chinese and English and still images. Alibaba’s executives bragged that the company has some of the most advanced diffusion transformer (DiT) architecture, enabling it to make 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 they envision users employing the AI video maker, but the company’s emphasis on third-party partnerships is suggestive. The technology might be employed in a range of marketing and entertainment videos. It might also end up in video games, producing visual references or even entire introductory videos.
Seen Sora?
The sheer number of AI video generators out or coming soon is astonishing, considering there weren’t any at the consumer level not long ago. OpenAI drew a lot of attention to the idea with its Sora model. Still, the company’s decision to limit Sora to certain partners left a lot of people hunting for alternatives, and companies like Alibaba are happy to fill in some of the gaps.
Runway, Stability AI, Pika, Hotshot, and Luma Labs’ Dream Machine are only some of the most prominent examples. And Alibaba isn’t alone among Chinese competitors. Kling and TikTok owner Bytedance’s Jimeng are in the same race. Alibaba has said action, but the final cut with final winners and losers has yet to be filmed.