TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Genie: Generative Interactive Environments

82 点作者 kuter大约 1 年前

7 条评论

jasonjmcghee大约 1 年前
&gt; Genie is capable of converting a variety of different prompts into interactive, playable environments that can be easily created, stepped into, and explored<p>If these are generating a fully interactive environments, why are all the clips ~1 second long?<p>Based on the first sentence in your paper, I would have expected a playable example as a demo. Or 20.<p>But reading a bit further into the paper, it sounds like the model needs to be actively running inference and will generate the next frame on the fly as actions are taken- is that correct?
评论 #39513643 未加载
polygamous_bat大约 1 年前
Firstly, do these models learn a good physics grounding for nonsense actions? Like keep pressing down even when you are in the ground? Or will they phase you through the ground?<p>Secondly, why are all videos like half a second long? I thought video generation came much farther than this. My guess would be that the world models unravel at any length longer than that, which is (and has always been) the problem with models such as these. Minus the video generation part, we had pretty good world models for games already, see Dreamer line of work: <a href="https:&#x2F;&#x2F;danijar.com&#x2F;project&#x2F;dreamerv3&#x2F;" rel="nofollow">https:&#x2F;&#x2F;danijar.com&#x2F;project&#x2F;dreamerv3&#x2F;</a>
评论 #39511958 未加载
nycdatasci大约 1 年前
The results seem quite bad. Compare the static image and &quot;game&quot; in this one example: Static Image: <a href="https:&#x2F;&#x2F;lh3.googleusercontent.com&#x2F;c0GV4hG0Xg0eqpsUS1z62v6aJ2qRGYKmmyJLbpfp1DbJ92elZ_lAvQpx4WSBVKbDO97OBvc0rhu_gNhXbzI_Uv121_QkX9Ur6m3MhMINg8kJ35RMNlA2lxQ2fIuennhBaQ=w1280" rel="nofollow">https:&#x2F;&#x2F;lh3.googleusercontent.com&#x2F;c0GV4hG0Xg0eqpsUS1z62v6aJ2...</a> &quot;Game&quot;: <a href="https:&#x2F;&#x2F;lh5.googleusercontent.com&#x2F;L_WsAa1saPmj29DSKda_fzk15y2Qk2Rvh3-b-4-n6EdXgaMJ6DrF5Chp3Oh9Nd6Pzz9IfmEwRJnM1t18zo2bsS9nNxpY1sr3pVJabbM-n1jf2bNbVAX9QEQiKlfcNYoQWQ=w1280" rel="nofollow">https:&#x2F;&#x2F;lh5.googleusercontent.com&#x2F;L_WsAa1saPmj29DSKda_fzk15y...</a><p>In the video, the character becomes a pixelated mess. In the static image, the character is clearly on rocks in the foreground, but in the &quot;game&quot; we see the character magically jumping from the foreground rocks to the background structure which also contains significant distortions.<p>The extremely short demo videos make it slightly harder to catch these obvious issues.
评论 #39511517 未加载
sqreept大约 1 年前
I&#x27;ve read twice the announcement and I can&#x27;t tell what this is good for. Can you please dumb it down for me?
snide大约 1 年前
I&#x27;m old an immediately assumed this would link to historical retrospective of GEnie<p><a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;GEnie" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;GEnie</a>
mdrzn大约 1 年前
Seems very interesting, but as soon as I see &quot;Google Research&quot; or &quot;Deepmind&quot; now it&#x27;s an instant turndown. Too much PR, not enough substance. Not targeting directly you guys with this research, but the company you work for.
joloooo大约 1 年前
Looking forward to following your progress. I&#x27;ve been wanting to see how we might replace polygons for gaming long term, this seems like a step in the right direction.