{"id":134267,"date":"2024-12-07T07:53:21","date_gmt":"2024-12-07T00:53:21","guid":{"rendered":"https:\/\/hotvideos24.online\/?p=134267"},"modified":"2024-12-07T07:53:21","modified_gmt":"2024-12-07T00:53:21","slug":"googles-genie-2-world-model-reveal-leaves-more-questions-than-answers","status":"publish","type":"post","link":"https:\/\/hotvideos24.online\/?p=134267","title":{"rendered":"Google\u2019s Genie 2 \u201cworld model\u201d reveal leaves more questions than answers"},"content":{"rendered":"<p> <script async src=\"https:\/\/pagead2.googlesyndication.com\/pagead\/js\/adsbygoogle.js?client=ca-pub-3711241968723425\"\r\n     crossorigin=\"anonymous\"><\/script>\r\n<ins class=\"adsbygoogle\"\r\n     style=\"display:block\"\r\n     data-ad-format=\"fluid\"\r\n     data-ad-layout-key=\"-fb+5w+4e-db+86\"\r\n     data-ad-client=\"ca-pub-3711241968723425\"\r\n     data-ad-slot=\"7910942971\"><\/ins>\r\n<script>\r\n     (adsbygoogle = window.adsbygoogle || []).push({});\r\n<\/script><br \/>\n<\/p>\n<div>\n<p>As podcaster Ryan Zhao <a href=\"https:\/\/bsky.app\/profile\/insrtcoins.bsky.social\/post\/3lcl6epzwm22k\">put it on Bluesky<\/a>, &#8220;The design process has gone wrong when what you need to prototype is &#8216;what if there was a space.'&#8221;<\/p>\n<h2>Gotta go fast<\/h2>\n<p>When Google revealed the first version of Genie earlier this year, it also <a href=\"https:\/\/arxiv.org\/pdf\/2402.15391v1\">published a detailed research paper<\/a> outlining the specific steps taken behind the scenes to train the model and how that model generated interactive videos. No such research paper has been published detailing Genie 2&#8217;s process, leaving us guessing at some important details.<\/p>\n<p>One of the most important of these details is model speed. The first Genie model generated its world at roughly one frame per second, a rate that was orders of magnitude slower than would be tolerably playable in real time. For Genie 2, Google only says that &#8220;the samples in this blog post are generated by an undistilled base model, to show what is possible. We can play a distilled version in real-time with a reduction in quality of the outputs.&#8221;<\/p>\n<p>Reading between the lines, it sounds like the full version of Genie 2 operates at something well below the real-time interactions implied by those flashy GIFs. It&#8217;s unclear how much &#8220;reduction in quality&#8221; is necessary to get a diluted version of the model to real-time controls, but given the lack of examples presented by Google, we have to assume that reduction is significant.<\/p>\n<figure class=\"ars-wp-img-shortcode id-2065422 align-center\">\n<div>\n                        <img width=\"800\" height=\"800\" src=\"https:\/\/cdn.arstechnica.net\/wp-content\/uploads\/2024\/12\/oasis.png\" class=\"center large\" alt=\"\" decoding=\"async\" loading=\"lazy\" srcset=\"https:\/\/cdn.arstechnica.net\/wp-content\/uploads\/2024\/12\/oasis.png 800w, https:\/\/cdn.arstechnica.net\/wp-content\/uploads\/2024\/12\/oasis-640x640.png 640w, https:\/\/cdn.arstechnica.net\/wp-content\/uploads\/2024\/12\/oasis-300x300.png 300w, https:\/\/cdn.arstechnica.net\/wp-content\/uploads\/2024\/12\/oasis-768x768.png 768w, https:\/\/cdn.arstechnica.net\/wp-content\/uploads\/2024\/12\/oasis-500x500.png 500w\" sizes=\"auto, (max-width: 800px) 100vw, 800px\"\/>\n                  <\/div><figcaption>\n<div class=\"caption font-impact dusk:text-gray-300 mb-4 mt-2 inline-flex flex-row items-stretch gap-1 text-base leading-tight text-gray-400 dark:text-gray-300\">\n<div class=\"caption-content\">\n      Oasis&#8217; AI-generated <em>Minecraft<\/em> clone shows great potential, but still has a lot of rough edges, so to speak.<\/p>\n<p>              <span class=\"caption-credit mt-2 text-xs\"><br \/>\n          Credit:<\/p>\n<p>                      <a class=\"caption-credit-link text-gray-400 no-underline hover:text-gray-500\" href=\"https:\/\/oasis-model.github.io\/\" target=\"_blank\" rel=\"noopener\"><\/p>\n<p>          Oasis<\/p>\n<p>                      <\/a><br \/>\n                  <\/span>\n          <\/div>\n<\/p><\/div>\n<\/figcaption><\/figure>\n<p>Real-time, interactive AI video generation isn&#8217;t exactly a pipe dream. Earlier this year, AI model maker <a href=\"https:\/\/www.decart.ai\/\">Decart<\/a> and hardware maker <a href=\"https:\/\/www.etched.com\/\">Etched<\/a> published <a href=\"https:\/\/oasis-model.github.io\/\">the Oasis model<\/a>, showing off a human-controllable, AI-generated video clone of <em>Minecraft<\/em> that runs at a full 20 frames per second. However, that 500 million parameter model was trained on millions of hours of footage of a single, relatively simple game, and focused exclusively on the limited set of actions and environmental designs inherent to that game.<\/p>\n<p>When Oasis launched, its creators fully admitted the model &#8220;struggles with domain generalization,&#8221; showing how &#8220;realistic&#8221; starting scenes <a href=\"https:\/\/oasis-model.github.io\/colloseum.webp\">had to be reduced to simplistic <em>Minecraft<\/em> blocks<\/a> to achieve good results. And even with those limitations, it&#8217;s not hard to <a href=\"https:\/\/www.forbes.com\/sites\/danidiplacido\/2024\/11\/03\/minecraft-is-finally-haunted-thanks-to-generative-ai\/\">find footage<\/a> of Oasis <a href=\"https:\/\/www.youtube.com\/watch?v=pWh4u2sXBhU\">degenerating into horrifying nightmare fuel<\/a> after just a few minutes of play.<\/p>\n<\/p><\/div>\n<p><script async src=\"https:\/\/pagead2.googlesyndication.com\/pagead\/js\/adsbygoogle.js?client=ca-pub-3711241968723425\"\r\n     crossorigin=\"anonymous\"><\/script>\r\n<ins class=\"adsbygoogle\"\r\n     style=\"display:block\"\r\n     data-ad-format=\"fluid\"\r\n     data-ad-layout-key=\"-fb+5w+4e-db+86\"\r\n     data-ad-client=\"ca-pub-3711241968723425\"\r\n     data-ad-slot=\"7910942971\"><\/ins>\r\n<script>\r\n     (adsbygoogle = window.adsbygoogle || []).push({});\r\n<\/script><br \/>\n<br \/><div data-type=\"_mgwidget\" data-widget-id=\"1660802\">\r\n<\/div>\r\n<script>(function(w,q){w[q]=w[q]||[];w[q].push([\"_mgc.load\"])})(window,\"_mgq\");\r\n<\/script>\r\n<br \/>\n<br \/><a href=\"https:\/\/arstechnica.com\/ai\/2024\/12\/googles-genie-2-world-model-reveal-leaves-more-questions-than-answers\/\">Source link <\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>As podcaster Ryan Zhao put it on Bluesky, &#8220;The design process has gone wrong when what you need to prototype is &#8216;what if there was a space.&#8217;&#8221; Gotta go fast &hellip; <a href=\"https:\/\/hotvideos24.online\/?p=134267\" class=\"more-link\">Read More<\/a><\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8630],"tags":[],"class_list":["post-134267","post","type-post","status-publish","format-standard","hentry","category-technology","entry"],"_links":{"self":[{"href":"https:\/\/hotvideos24.online\/index.php?rest_route=\/wp\/v2\/posts\/134267","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/hotvideos24.online\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/hotvideos24.online\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/hotvideos24.online\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/hotvideos24.online\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=134267"}],"version-history":[{"count":0,"href":"https:\/\/hotvideos24.online\/index.php?rest_route=\/wp\/v2\/posts\/134267\/revisions"}],"wp:attachment":[{"href":"https:\/\/hotvideos24.online\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=134267"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/hotvideos24.online\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=134267"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/hotvideos24.online\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=134267"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}