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。

MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Images

131 点作者 jasondavies10 个月前

5 条评论

axoltl9 个月前
I&#x27;m having a hard time finding a reference to the hardware the inference is run on. The paper mentions training was done on a single A100 GPU so I&#x27;m going to assume inference was run on that same platform. The 22fps result is somewhat meaningless without that information.<p>It does feel like we&#x27;re getting closer and closer to being able to synthesize novel views in realtime from a small set of images at a framerate and quality high enough for use in AR, which is an interesting concept. I&#x27;d love to be able to &#x27;walk around&#x27; in my photo library.
评论 #41243356 未加载
评论 #41243297 未加载
vessenes9 个月前
The tech stack in the splat world is still really young. For instance, I was thinking to myself: “Cool, MVSplat is pretty fast. Maybe I’ll use it to get some renderings of a field by my house.”<p>As far as I can tell, I will need to offer a bunch of photographs with camera pose data added — okay, fair enough, the splat architecture exists to generate splats.<p>Now, what’s the best way to get camera pose data from arbitrary outdoor photos? … Cue a long wrangle through multiple papers. Maybe, as of today… FAR? (<a href="https:&#x2F;&#x2F;crockwell.github.io&#x2F;far&#x2F;" rel="nofollow">https:&#x2F;&#x2F;crockwell.github.io&#x2F;far&#x2F;</a>). That claims up to 80% pose accuracy depending on source data.<p>I have no idea how MVSplat will deal with 80% accurate camera pose data… And I also don’t understand if I should use a pre-trained model from them or train my own or fine tune one of their models on my photos… This is sounding like a long project.<p>I don’t say this to complain, only to note where the edges are right now, and think about the commercialization gap. There are iPhone apps that will get (shitty) splats together for you right now, and there are higher end commercial projects like Skydio that will work with a drone to fill in a three dimensional representation of an object (or maybe some land, not sure about the outdoor support), but those are like multiple thousand-dollar per month subscriptions + hardware as far as I can tell.<p>Anyway, interesting. I expect that over the next few years we’ll have push button stacks based on ‘good enough’ open models, and those will iterate and go through cycles of being upsold &#x2F; improved &#x2F; etc. We are still a ways away from a trawl through an iPhone&#x2F;gphoto library and a “hey, I made some environments for you!” Type of feature. But not infinitely far away.
评论 #41250725 未加载
评论 #41248211 未加载
petargyurov9 个月前
Someone help me understand inference here.<p>Every gaussian splat repo I have looked at doesn&#x27;t mention how to use the pre-trained models to &quot;simply&quot; take MY images as input and output a GS. They all talk about evaluation, but the CMD interface requires the eval datasets as input.<p>Is training&#x2F;fine-tuning on my data the only way to get the output?
评论 #41244238 未加载
评论 #41244392 未加载
programjames9 个月前
Where would you use 3D Gaussian splatting? Static environments for video games?
评论 #41242604 未加载
评论 #41244294 未加载
评论 #41246356 未加载
评论 #41243083 未加载
评论 #41242865 未加载
评论 #41244411 未加载
评论 #41242437 未加载
评论 #41242820 未加载
评论 #41246634 未加载
评论 #41248290 未加载
评论 #41243328 未加载
评论 #41242283 未加载
评论 #41242145 未加载
rebuilder9 个月前
The indoor example with the staircase and railing was really surprising - there&#x27;s only one view of much of what&#x27;s behind the doorframe and it still seems to reconstruct a pretty good 3d scene there.