2024-10-07 08:38:40
Meta presented a series of fundamental models that it collects under the name Movie Gen, with which it offers creators tools to generate, customize and edit hyper-realistic videos also with audio through textual descriptions. In this way, Zuckerberg joins other companies in creating artificial intelligence (AI) tools to generate videos, as OpenAI did with Sora last February.
This new tool, in Meta’s wordsis aimed at content creators and filmmakers, with the aim of helping “enhance their creativity, rather than replace it”. Movie Gen has two functional models, one aimed at video (Movie Gen Video), with 30,000 million parameters, and another focused on sound generation (Movie Gen Audio), with 13,000 million parameters.
Zuckerberg puts it to the test
As Meta explains, the way Movie Gen works is quite similar to other utilities of this type. With just the textual description it is possible to create a video lasting between 4 and 16 seconds, at 16 frames per second. The AI also allows you to edit existing clips using different text descriptions or even create custom videos by uploading a photo of the user. Although the company claims that the material is hyper-realistic and has Full HD quality, it is surprising that Meta chose to make them at 16 FPS and not in a 24 frames per second style, as is done in the film industry.
Zuckerberg himself provided a first look at Meta Movie Gen’s capabilities through an Instagram post. In it you can see him exercising, while different elements of the background, his clothes or devices change depending on what the artificial intelligence is asked.
Ability to generate audio
One of Movie Gen’s key differentiators is its ability to generate sounds for the videos in question. Keep in mind that tools like Sora, for example, do not offer this possibility. According to its creators, the 13 billion parameter model can use a video and text description to generate an audio track that matches what’s happening in the image.
Among the examples Meta shared, a quad bike is seen speeding up and jumping, with the engine noise playing in the background along with the music. You can also see a snake moving through the vegetation, with the noise of the leaves and the corresponding musical accompaniment also created with artificial intelligence. On this occasion, the audio allows a duration of up to 45 seconds and can reach anything from ambient sounds to instrumental music. However, it does not allow you to generate voices or dialogues, probably to avoid the generation of deepfakes.
#Meta #rivals #Sora #Movie #Gen #generates #videos