Google shared video clips it said were generated by Veo, pictured above
| Photo Credit: Google
Google has introduced what it called its “most capable” video generation model yet, named Veo, which it said would generate high-quality, 1080p resolution videos that are over 60 seconds long, in a variety of styles.
Veo has not been publicly released, but a limited number of creators will be able to try it out through the experimental VideoFX tool, while other interested applicants can join a waitlist.
Google shared video clips it said were generated by Veo, which included hyper-realistic aerial nature shots as well as whimsical clips such as a crocheted elephant walking and alpacas dancing.
There were not many clips that showed humans or their faces and bodies clearly. Some clips had a slightly over-edited feel, pointing to their AI origins.
Google agreed that maintaining visual consistency in such cases could be a challenge, but said that filmmakers and creators will be sharing their feedback regarding Veo.
(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)
“It accurately captures the nuance and tone of a prompt, and provides an unprecedented level of creative control — understanding prompts for all kinds of cinematic effects, like time lapses or aerial shots of a landscape,” said Google in its post.
Earlier this year, ChatGPT-maker OpenAI released its Sora text-to-video generator to a limited team of testers.
OpenAI shared sample clips that it said Sora had generated, such as a female model walking down a neon-soaked city street. However, issues such as warped movements and body parts clearly showed that the clips were generated with AI.
OpenAI CTO Mira Murati was also unable to answer questions about whether the company had trained its video generator with YouTube videos.
Google said that Veo was responsible by design.
“Videos created by Veo are watermarked using SynthID, our cutting-edge tool for watermarking and identifying AI-generated content, and passed through safety filters and memorization checking processes that help mitigate privacy, copyright and bias risks,” said the tech giant.
