Show HN: Vidu – The New text/image-to-video MultiModal LLM

Hi HN, Vidu team here. We're excited to introduce our latest project, presented by Tsinghua University and Shengshu Technology - Vidu, a text and image-to-video MultiModal LLM. Professor Jun Zhu and his colleagues published a research paper on Vidu and its capabilities during Vidu's early stages: https://arxiv.org/abs/2405.04233 https://arxiv.org/abs/2405.04233 To set Vidu apart from the mass generative video AI that has already been released, our team chose to focus on augmenting specific properties during the early-stages of Vidu's developments, such as consistency. Inconsistencies has to be the most expressed concern regarding generative AI. So one of our key goals is to achieve a level of universal consistency rarely seen in generative video AI. This means that, regardless of camera angle changes, the video subject—whether a person or an object—remains consistent and coherent. Through our team working endlessly on iterations and refining, Vidu is now capable of generating exceptional videos of many styles within the shortest amount of time. We are especially proud to present Vidu's superior generation capability. We are confident that Vidu has the ability to maintain animation consistency and generate the best-quality animations. Besides excellent animation generations, Vidu's features now consist: - Precise comprehension of prompt text - Dynamic motions and visual impact - Proficiency in recreating various movie/ art styles - Capable of diverse camera angles - Realistic VFX - Character to Video We appreciate any feedback, suggestions, or even critiques! Send your comments our way!