Google debuts new AI video generator Veo 2 claiming better audience scores than Sora

Google Confronting head-to-head Open artificial intelligenceSora and its latest version of the video generation model, Veo 2, it says make Videos that look more real.

The company also updated its image generation model, Imagen 3, to produce richer, more detailed photos.

Google says Veo 2 “can better understand real-world physics and the nuances of human movement and expression.” It can be used Google Labs Video special effects platform – But only on the waiting list. Users need to register via a Google form and wait for Google to temporarily grant access at a time of their choosing.

“Veo 2 also understands the language of cinematography: asking for genres, specifying lenses, and suggesting cinematic effects, Veo 2 will deliver resolutions up to 4K,” Google said in a blog post.

While Veo 2 is only available to some users, the original Veo remains available About Vertex AI. Videos created with Veo 2 will contain Google’s metadata watermark SynthID to identify them as AI-generated.

Google admits, however, that the Veo 2 may still produce hallucinations such as extra fingers, but it promises the new model will produce fewer hallucinations.

Veo 2 will compete with OpenAI’s recently released Sora video generation model to attract filmmakers and content creators. Sora had been previewed for some time before OpenAI was released available To paying subscribers.

Impressively, Google says that in its internal tests that measure “overall preference” (i.e. which videos viewers prefer) and “timely compliance” (how well a video matches the instructions given by its human creator), Veo Sora and other competitor AI models are favored by human evaluators.

Google announced I saw in possible During this year’s Google I/O developer conference, we collaborated on a video with actor and musician Donald Glover (aka Childish Gambino).

AI video generation still needs work

Artificial intelligence video generation has long been an area of ​​generative artificial intelligence, with large model developers such as Google and OpenAI often competing and catching up with relatively smaller companies.

runway machine learningis one of the pioneers in artificial intelligence video generation and recently launched advanced controls for it Gen-3 Alpha Turbo model. pickup lab Release Pickup 2.0giving users more control so they can add their own characters to the movie. forward artificial intelligence Announcing partnership with AWS Bring its model to bedrock For corporate use. Lumaye Expand its dream machine Generation model.

However, AI video generation still requires convincing creators and audiences. After Sora’s long-awaited release, People are still skeptical Its capabilities were demonstrated as it continued to generate figures that defied physics and anatomy. Users feel it gives inconsistent results.

The recent Game Awards trailer also show people’s distrust They think “artificial intelligence is bullshit.”

However, some filmmakers have begun embrace possibilities AI video generator can provide. Renowned director James Cameron joins board of directors Stable artificial intelligencewhile actor Andy Serkis announced that he is setting up a production company focused on artificial intelligence.

Google says it’s seeing a lot of user interest. YouTube creators have been using VideoFX to create backgrounds for YouTube Shorts to save time, the company said.

Imagen 3 Update

Google has also updated its image model Imagen 3, recently made available via the Gemini chatbot on the web, to be more realistic and provide brighter images.

Imagen 3 can now accurately render more art styles, “from photorealism to impressionism, and abstraction to animation.” Google says the model will also follow cues more faithfully.

Imagen 3 can be accessed through ImageFX.