Sign in to your Android Police account
Summary
- Google’s Veo 2 AI video generator creates highly realistic videos with minimal flaws or hallucinations.
- Veo 2 outputs videos with a SynthID watermark to combat misinformation and identify them as AI-generated.
- Veo 2 features improvements in realism, physics, cinematic effects, resolution, and longer clip outputs compared to the original.
OpenAI brought us Sora, an AI-powered video-creation tool that makes incredibly lifelike videos. Not to be outdone, Google just launched Veo 2, the latest model of its video generator. Google says it creates highly realistic videos and addresses flaws like hallucinations.
Related
Google Veo: The ultrarealistic AI video generation tool explained
Close-ups, panning, and all the film-school tricks at your fingertips
Like the original Veo, the new Veo 2 outputs videos with an invisible SynthID watermark so they are identifiable as AI-generated. This helps combat misinformation, at least on paper.
Everything new in Veo 2
Google released a short video, created with Veo 2, to highlight how far it has come. It is a shockingly realistic video and shows a woman looking into a microscope, a cartoon child in a kitchen, and a dog swimming in a pool. All of these images look miles ahead of what the original Veo could produce.
Google said Veo 2 includes significant improvements in key areas:
Realism and physics: Veo 2 has better understanding of human movement, facial expressions, and real-world physics.
Cinematic improvements: Users can specify shot types, lenses, and cinematic effects such as wide-angle or shallow depth of field.
Few hallucinations: Google claims Veo 2 reduces the number of weird things we’ve come to expect from AI videos, such as extra fingers or distorted faces.
Higher resolution: Veo 2 can produce videos in up to 4K resolution.
Longer outputs: The updated video generator supports clips of several minutes in length.
Part of Google’s drive to become the biggest AI company
Veo 2 complements Google’s bigger picture as the company pivots from being an internet company to an AI company. Google also announced a new Imagen 3 AI image generator, and Whisk, a tool that lets users drop multiple images in as prompts and will create a brand new image.
Veo 2 is available now for select users via the Google Labs VideoFX platform. There’s also a waiting list for the rest of the US public who want to try their hand at making a minute-long short video.
[ For more curated Samsung news, check out the main news page here]
The post Google says Veo 2 AI can generate videos without all the hallucinations first appeared on www.androidpolice.com