Recently, a generative AI company incubated by Y-CombinatorMoonvalley.ai has launched a powerful generative text-to-video model that can generate HD, 16:9 film-quality videos using only text prompts, with the ability to choose from a variety of video styles such as surrealism, anime, and fantasy.
The product is available in both web and APP versions. Users can easily generate videos through Discord channels, just as easily as creating images in Midjourney, but the video quality is greatly improved. Moonvalley said that they have invested all their time and resources to solve the problem that the progress of text-to-video generation AI in the market is not fast enough. The company's mission is to "turn text into the world" and has been supported by well-known institutions such as Y-Combinator and Khosla Ventures.
Moonvalley claims that this is the most powerful AI video generation engine ever. Currently, the model is in the Beta stage, completely free, and uses Discord as the UI. New users need to register and log in to the community interface before they can experience this new tool.
Users can also add lens operation methods in Prompt to make the video expression more personalized. However, compared with other text-to-video generation applications, Moonvalley is slower and requires waiting for a while to get results. Since it is difficult for an input to meet expectations, it often requires repeated modifications. If you have to wait for more than ten minutes or even longer each time, it may affect the user's interest.
In addition, some netizens have reported that the model needs to be trained specifically for the tiny details of the hands and feet. For example, most people are born with 10 fingers and 10 toes, with few exceptions, but these common sense errors still appear in the works generated by the model. The solution to this problem is to use negative prompts.