Meta has just presented Movie Gen, its video creation application from text or image. Many features, which will be found on the group's platforms. For now, it is in the testing phase.

A photo, a text, and a simple prompt to transform them into a creative video, add a skater in action to a video and then add a red cape in one click… That's it, Meta has done it! The Californian firm created by Mark Zuckerberg has crossed a new step in the field of generative artificial intelligence with the presentation of Movie Gen, its new AI model capable of creating videos from simple text or visual queries. What are now called “Text to video” apps.

Unveiled this Friday, October 4 in New York, It is part of a series of innovations launched by Meta in the generation of multimedia content., following two previous models dedicated to image generation, released in 2022 and 2023. Apparently very efficient, powerful and easy to use, Movie Gen, which is arriving on an already busy market (Synthesia, Sora, Lumen 5, Imagen, Phenaki, etc.), will not be immediately available to the general public; it is currently being tested by a select group of creators and directors, with the aim of improving its features before its release on the market.

Generating videos from text and images

Movie Gen's main feature is its ability to generate videos from a simple text query (prompt). By entering a description in plain language, the user can get a video generated entirely by AI. This approach is similar to that used for AI image generation, but extends the complexity to full audiovisual content.

Additionally, Movie Gen allows you to create videos from a single image. This means that even a simple photo can be transformed into an animated sequence, offering impressive creative possibilities.

Editing and enriching existing videos

In addition to the ex nihilo generation, Meta offers with Movie Gen a function of editing pre-existing videos. Users can thus modify or enrich a video based on instructions formulated in natural language. This ability to edit video footage, whether for adjusting visual elements or integrating new effects is one of the strengths of this interface. For example, it would be possible to change the weather or mood of a scene simply by describing the desired change.

Adding soundtrack

Another key feature of Movie Gen is its ability to add a soundtrack to generated videos. As with visual elements, simply make a request and the AI will add music or sound effects to the footage. In a demo published by Meta, A simple sentence accompanied the video of a quad driving in the desert with the sound of the engine and a guitar soundtrack. This ability to synchronize image and sound expands the creative possibilities and greatly simplifies the video creation process for users.

Integration prospects on Meta platforms

While Movie Gen remains in the testing phase for now, Meta hints at future applications on its own platforms. It is therefore possible that this AI will be integrated into social networks such as Instagram and Facebook, or even into messaging tools such as WhatsApp., platforms owned by Meta. For example, a user could edit a video via Movie Gen before publishing it directly to their networks, all from instructions in plain language. This potential integration could transform the way videos are created and shared on social networks, offering users a powerful tool to enrich their content without requiring advanced technical skills. With the risk of an overproduction of video and the increase in the time spent scrolling on networks…

Comparison with other video AI players

Meta isn't the first company to get into AI video generation. Runway, for example, a New York startup specializing in artificial intelligence, offers Gen-3, a similar model that can create short videos from text queries or transform a series of still images into animated sequences, change backgrounds, etc. Similarly, OpenAI launched its own video model in February 2024, called Sora, and Google unveiled it a month earlier Light. We must take into account PixVerse, Creatus AI and Luma Dream Machine.  Despite this competition, Meta claims that Movie Gen "outperforms" these competitors, especially when the quality of the videos is evaluated by humans., explained the promoters at Meta when presenting Movie Gen.

This assurance shows the social media giant's confidence in its model's ability to establish itself – and for good reason! – as a benchmark in the field of AI video generation.