Vidu 1.5 — это амбициозный ответ китайских разработчиков модели Sora от OpenAI. Нейросеть мастерски справляется с генерацией видео по тексту или изображению, предлагая новый уровень качества визуализации и динамики.
Shengshu Technology has revealed its latest AI model, the Vidu 1.5. The company aims to venture into the domain of text-to-video and image-to-video generation, bringing a new competitor into the market to compete against the likes of OpenAI Sora, which was announced earlier this year. For someone who is not aware of Shengshu Technology, it’s an emerging AI company that was founded last year in March 2023. The company’s new AI model is pretty much similar to OpenAI’s Sora, albeit with some tweaks. The Vidu 1.5 has the capability to generate videos up to eight seconds in length from images as well as textual prompts. The AI model uses its in-house multiple-entity consistency feature to seamlessly splice people, objects, and environments while generating a video from user prompts. There’s also a thing that Vidu called multiple-angle consistency, which allows users to either generate videos using any inputted images or by uploading three photos of a single subject. The AI firm further states that the AI model utilizes advanced control features for adding better motion and detailed backgrounds in the generated output. According to the model maker, you can generate a eight-second video in under 30 seconds.