{"id":24458,"date":"2024-12-05T09:33:49","date_gmt":"2024-12-05T01:33:49","guid":{"rendered":"https:\/\/www.1ai.net\/?p=24458"},"modified":"2024-12-05T09:33:49","modified_gmt":"2024-12-05T01:33:49","slug":"%e8%b0%b7%e6%ad%8c%e6%97%97%e4%b8%8b-deepmind-%e6%8e%a8%e5%87%ba-genie-2-%e6%a8%a1%e5%9e%8b%ef%bc%8c%e5%8f%af%e7%94%9f%e6%88%90%e9%95%bf%e8%be%be-1-%e5%88%86%e9%92%9f%e7%9a%84%e6%b8%b8%e6%88%8f","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/24458.html","title":{"rendered":"Google's DeepMind Launches Genie 2 Model to Generate Game Worlds Up to 1 Minute Long"},"content":{"rendered":"

December 5 News.Google<\/a>Its Artificial Intelligence Research Organization DeepMind<\/a> A new model called Genie 2 has been released that generates an \"infinite\" variety of playable 3D worlds from a single image and text description. An update to the Genie model introduced earlier this year, Genie 2 marks a major breakthrough for artificial intelligence in virtual world generation.<\/p>\n

\"df747020j00snzyz80072d000hs00a0p\"<\/p>\n

Genie 2 is capable of generating interactive 3D scenes in real time based on textual descriptions and images entered by the user.<\/strong>For example, enter \"cute humanoid robot in the forest\" and the model will build a dynamic scene with a robot character and an explorable environment. For example, by typing \"cute humanoid robot in the forest\", the model builds a dynamic scene with a robot character and an explorable environment. The user can interact with the character by using the keyboard or mouse to jump, swim, etc. in the world.<\/p>\n

DeepMind says Genie 2 can generate coherent worlds with different perspectives (e.g., first-person and isometric).They last up to a minute, with most lasting 10 to 20 seconds.<\/strong><\/p>\n

\"3469ba2cj00snzyyq009od000hs00a0m\"<\/p>\n

DeepMind also says that Genie 2 is able to simulate object interaction, animation, lighting, physical reflections, and the behavior of \"non-player characters\" (NPCs) during the generation process. Many of the scenes generated are close to AAA electronic quality.game<\/a>, and even excels in object viewpoint consistency and scene memory.<\/p>\n

Similar models are available from World Labs, founded by Fei-Fei Li, and Israeli startup Decart, and while most models like Genie 2 -- a.k.a. the World Model -- can simulate games and 3D environments, they suffer from artifacts and illusion-related problems. But there are problems related to artifacts, consistency and illusion. For example, Decart's Minecraft simulator, Oasis, is low-resolution and quickly \"forgets\" level layouts. Genie 2, however, remembers parts of the simulated scene that are out of view and renders them accurately when they become visible again. (World Labs' models can do the same.)<\/p>\n

It's worth noting that DeepMind did not disclose the source of Genie 2's training data in detail, but industry speculation suggests that it may include playthroughs of a large number of popular games. Considering that Google has access to YouTube's vast video resources and claims the right to use its content for training, this has sparked controversy over whether the model infringes on intellectual property rights.<\/p>\n

Games currently created with Genie 2 won't actually be that interesting, as progress is erased every minute or so. As a result, DeepMind is positioning it as a research and creativity tool for scenarios such as rapid prototyping and AI intelligences evaluation.<\/p>\n

In its blog post, DeepMind writes, \"With Genie 2's generalization capabilities, concept art and hand-drawn sketches can be transformed into fully interactive environments. This allows researchers to quickly generate diverse environments to support evaluation of unseen task scenarios.\"<\/p>\n

1AI notes that Google's investment in world modeling research continues to expand. In October, DeepMind hired Tim Brooks, the former head of OpenAI's video generation program, while two years ago it poached Tim Rockt\u00e4schel, known for his open-ended experiments, from Meta.<\/p>","protected":false},"excerpt":{"rendered":"

On December 5th, DeepMind, a Google-based artificial intelligence research institute, published a new model called Genee 2, which can generate an \u201cinfinite\u201d 3D world with a single picture and text description. As an upgraded version of the Genie model, launched early this year, Genie 2 marked a major breakthrough in the field of artificial intelligence generation in the virtual world. Genie 2 can generate interactive three-dimensional scenarios in real time, based on text descriptions and images entered by the user. For example, the input of a \u201cloved human robot in the forest\u201d model can create a dynamic scene that contains robotic roles and explores the environment. Users can interact in the world by jumping, swimming, etc. through keyboard or mouse-operator roles. DeepM<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[593,1409,281],"collection":[],"class_list":["post-24458","post","type-post","status-publish","format-standard","hentry","category-news","tag-deepmind","tag-1409","tag-281"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/24458","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=24458"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/24458\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=24458"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=24458"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=24458"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=24458"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}