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。

GET3D: A Generative Model of High Quality 3D Textured Shapes Learned from Images

177 点作者 lnyan超过 2 年前

10 条评论

etaioinshrdlu超过 2 年前
In somewhat related topics, I think we can just use stable diffusion to help convert single photos to 3D NERFs.<p>1. find the prompt that best generates the image<p>2. generate a (crude) NERF from your starting image and render views from other angles<p>3. use stable diffusion with the views from other angles as seed images, refine them using the prompt from 1 combined with(add descriptions to generate &quot;view from back&quot;, &quot;view from top&quot;, etc<p>4. feed the refined views back to the NERF generator, keeping the initial photo view constant<p>5. Generate new views from the NERF, which should now be much more realistic.<p>Run the above steps 2-5 in a loop indefinitely. Eventually you should end up with a highly accurate, realistic NERF which is full 3d from any angle, all from a single photo.<p>Similar techniques could be used to extend the scene in all directions.
评论 #32966625 未加载
评论 #32968784 未加载
评论 #32965977 未加载
TOMDM超过 2 年前
Ok, so on the generative model modality landscape I&#x27;m now aware of:<p>- speech<p>- images<p>- audio samples<p>- text<p>- code<p>- 3d models<p>I&#x27;ve seen basic attempts at music and video, and based on everything else we&#x27;ve seen getting good results there seems to be mostly a matter of scaling.<p>What content generation modalities are left? Will all corporate generation of these fall to progressively larger models, leaving a (relatively) niche &quot;Made by humans!&quot; industry in it&#x27;s wake?
评论 #32963987 未加载
评论 #32963449 未加载
评论 #32966393 未加载
评论 #32971658 未加载
评论 #32966398 未加载
评论 #32963833 未加载
评论 #32963643 未加载
评论 #32966311 未加载
评论 #32963458 未加载
ummonk超过 2 年前
Still nowhere near good enough to be able to generate a VFX or video game asset from some pictures, which is what we&#x27;d really want for a practical application of such a tool.
评论 #32969694 未加载
评论 #32966893 未加载
calibas超过 2 年前
Some of the videos aren&#x27;t working in Firefox. Here&#x27;s the error:<p>&gt; Can&#x27;t decode H.264 stream because its resolution is out of the maximum limitation
评论 #32967053 未加载
bno1超过 2 年前
An AI that does good UV unwrapping would be much more interesting and useful.
评论 #32965476 未加载
评论 #32965366 未加载
评论 #32970092 未加载
incrudible超过 2 年前
Spoiler: The results are not high quality, at all.
评论 #32965639 未加载
评论 #32971190 未加载
jokethrowaway超过 2 年前
Great, now we can get the unreleased code for this paper and use it with the unreleased code for generating animations (really impressive stuff by Sebastian Starke, presented at various SIGGRAPH) and build a videogame generator.<p>I wouldn&#x27;t even mad if it were a paid product and not free code, just release something to the world so we can start using it.
nekopa超过 2 年前
One step closer to a dream I have: To describe scifi objects to Stable Diffusion, use the image to create a 3D object, print that object on my 3D printer. All on my laptop. (Well, I have SD running at home now, will have to see how the code for this runs when it is finally released)
评论 #32982200 未加载
corscans超过 2 年前
Hecking man
wokwokwok超过 2 年前
<a href="https:&#x2F;&#x2F;github.com&#x2F;nv-tlabs&#x2F;GET3D" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;nv-tlabs&#x2F;GET3D</a><p>&gt; News<p>&gt; 2022-09-22: Code will be uploaded next week!<p>Not really that interesting at this point; the 5 page paper has a lot of hand waving, and without the code to see how they actually implemented it…<p>…I’m left totally underwhelmed.<p>No weights.<p>No model.<p>No code.<p>The pictures were very pretty.<p>&#x2F;shrug
评论 #32964497 未加载
评论 #32963895 未加载
评论 #32964815 未加载