How it would look in reality.

現實中的樣子

That is called rendering.

這就是所謂的渲染。

One of my favorite ways to do rendering is through ray tracing, which is an incredible technique that simulates how light interacts with the scene and gives you these beautiful results.

我最喜歡的渲染方式之一是光線追蹤，這是一種令人難以置信的技術，它能模擬光線與場景的交互作用，為你帶來美麗的效果。

That was rendering.

這就是渲染。

Scene goes in, an image comes out.

場景進去，影像出來。

But now, imagine the opposite.

但現在，想象一下相反的情況。

What is this crazy talk?

這是什麼瘋話？

Well, hear me out.

好吧，聽我說完。

Imagine that you have images, and you want to get what the scene was behind the image.

想象一下，您有影像，而您想獲得影像背後的場景。

Get a digital 3D model of the scene.

獲取場景的數字 3D 模型。

An image goes in, and a scene comes out.

一個影像進去，一個場景出來。

And imagine how cool that would be when you want to create a video game, you just make an image, and out comes the scene for the game.

想象一下，當你想製作一款視頻遊戲時，只需製作一張圖片，遊戲場景就會出現，那將是多麼酷的一件事。

However, not so fast.

然而，沒那麼快。

Dear Fellow Scholars, this is Two Minute Papers with Dr. Károly Zsolnai-Fehér, and here, the question is, what is the geometry there, what materials are they made of, lighting everything?

親愛的學者朋友們，這裡是《兩分鐘論文》，請收看 Károly Zsolnai-Fehér 博士的文章，問題是，那裡的幾何形狀是怎樣的？

That needs an expert and many-many work hours.

這需要專家和大量的工作時間。

Here you see Andrew Price assembling such a scene in Blender, a 3D editor program.

在這裡，你可以看到安德魯-普萊斯在 Blender（一款三維編輯程序）中組裝這樣一個場景。

Imagine that you have a photo of the result, then, you have to do all this sculpting, assigning materials, lighting, and more.

試想一下，如果你有一張效果圖的照片，那麼，你就必須完成所有的雕刻、指定材料、照明等工作。

And then, when you are done, rendering.

完成後，再進行渲染。

But at this point, you just think that you are done, because your results are obviously not the same as the target image, so you have to play with the geometry, lighting, and materials, render the image again.

但此時，你會認為你已經完成了，因為你的結果顯然與目標影像不一樣，所以你必須對幾何圖形、光照和材質進行調整，再次渲染影像。

This takes forever.

這需要很長時間。

Hours.

工作時間

Days.

天。

Even weeks, depending on the complexity of the scene.

甚至幾周，取決於場景的複雜程度。

And it is super challenging.

而且它極具挑戰性。

So, we look at a photo and build a scene by hand.

是以，我們看著一張照片，然後手工構建一個場景。

I imagine that in a science fiction world, we could, perhaps, have an amazing algorithm that could do this automatically.

我想象，在科幻世界裡，我們或許可以擁有一種神奇的算法，可以自動完成這些工作。

Let's call this inverse rendering.

我們稱之為反向渲染。

How would that even work?

這怎麼行得通？

Well, few of you Fellow Scholars know, there are previous works that do that.

很少有研究員學者知道，以前也有這樣的作品。

They are absolutely amazing.

它們絕對令人驚歎。

You see how these beautiful pieces of geometry grow out of nothing.

你看，這些美麗的幾何圖形是如何無中生有的。

You just specify what you want with a 2D image, and it creates a 3D model.

你只需指定你想要的二維影像，它就會創建一個三維模型。

And we are talking detailed 3D models here.

我們現在討論的是詳細的 3D 模型。

Super incredible.

超級不可思議

But this is just 3D modeling.

但這只是 3D 建模。

What about materials?

材料如何？

And now, have a look at this research paper from the University of California, Irvine and NVIDIA.

現在，來看看加州大學歐文分校和英偉達公司的研究論文吧。

And it can do some amazing things.

它能做出一些驚人的事情。

First, if we put a few light sources on a painting, it can reconstruct the painting itself.

首先，如果我們在一幅畫上放置幾個光源，它就能重建畫作本身。

It is similarly good with reconstructing the material of this object too if it can take a look at a set of images of it.

如果它能查看該物體的一組影像，也同樣能重建該物體的材料。

But here come my favorites.

不過，我最喜歡的還是這裡。

Here is a tree.

這裡有一棵樹。

Now, hold on to your papers Fellow Scholars, because the task is super challenging.

現在，請拿好你們的論文 各位學者，因為這項任務極具挑戰性。

What you get is not even a view of the plant.

你看到的甚至不是工廠的景象。

No, no, you get a look at just the shadow of the plant.

不，不，你只能看到植物的影子。

Now, figure out what the plant looks like.

現在，弄清楚植物長什麼樣。

Previous techniques couldn't do it, they were just trying and trying, and not getting anywhere.

以前的技術無法做到這一點，他們只是在不斷嘗試，卻毫無進展。

But the new method, first, it tries to sculpt the object in different possible ways to be able to match its shadow, and as it does that, you can see its current guess for the geometry of the tree itself.

但新方法首先會嘗試以各種可能的方式雕刻對象，以匹配其陰影，在雕刻過程中，你可以看到它對樹本身幾何形狀的當前猜測。

So, can it do it?

那麼，它能做到嗎？

It can!

可以！

Wow!

哇

Absolutely amazing!

絕對令人驚歎！

And all this from just seeing a shadow.

而這一切，只是因為看到了一個影子。

The processes that you see here are sped up, and this one took only 16 minutes.

你在這裡看到的過程都被加速了，這個過程只用了 16 分鐘。

I think, for a human being, doing this sounds almost impossible in any amount of time, let alone in a few minutes.

我認為，對於一個人來說，在任何時間內做到這一點聽起來都幾乎是不可能的，更不用說在幾分鐘內了。

Incredible!

不可思議

Now, we have two more really cool tests for it, and then, a huge surprise!

現在，我們還要對它進行兩項非常酷的測試，然後是一個巨大的驚喜！

Here is one more case where we reconstruct an octagon by merely looking at its shadow again.

這裡還有一種情況，我們只需再次觀察八邊形的影子，就能重建八邊形。

This one checks out, and now, we have a world map relief in a large room with a window.

現在，我們在一個有窗戶的大房間裡看到了一幅世界地圖浮雕。

It gets to look at some images of it, but ultimately, what it needs to reconstruct is the bumps in the geometry that create this effect.

它可以查看一些影像，但最終需要重建的是產生這種效果的幾何圖形中的凹凸。

And I love seeing how it morphs this solution into being.

我喜歡看到它是如何將這一解決方案變為現實的。

So cool!

太酷了

And here comes the best part!

最精彩的部分來了

This technique is up to a hundred times faster than its predecessors.

這種技術比其前代產品快達 100 倍。

A hundred X!

100 X！

That is an incredible leap forward in just one research paper.

僅僅是一篇研究論文，就實現了令人難以置信的飛躍。

And since the quickest scene goes from 12 minutes up to about 2 hours, that can already be useful for some tasks, but just imagine what we will be capable of two more papers down the line!

由於最快的場景從 12 分鐘到約 2 小時不等，這對於某些任務來說已經很有用了，但試想一下，如果再多出兩篇論文，我們將能做些什麼！

My goodness!

我的天啊

So, we don't need to live in a science fiction world for a technique like this, it is right here!

是以，我們不需要生活在科幻世界中，這樣的技術就在我們身邊！

I am getting goosebumps.

我都起雞皮疙瘩了。

And it is clearly not as good as Andrew, no one is, but it is an amazing step forward in creating virtual worlds, maybe even video games from just an image or a drawing.

它顯然沒有安德魯那麼優秀，沒有人比他更優秀，但它在創建虛擬世界方面邁出了驚人的一步，甚至可以通過影像或繪畫來創建視頻遊戲。

Don't forget, scientists at Google DeepMind are already working on the video game part.

別忘了，谷歌 DeepMind 的科學家們已經在研究視頻遊戲的部分了。

And good news, the source code is also available.

好消息是，我們還提供了源代碼。

So we get all this knowledge for free!

是以，我們可以免費獲得這些知識！

I don't know if I told you before, but I love research papers.

不知道我以前有沒有告訴過你，我喜歡研究論文。

And if you wish to run your own experiments in the cloud, make sure to check out Microsoft

如果您希望在雲中運行自己的實驗，請務必查看微軟

Azure AI.

Azure AI。

And by the way, Azure AI is a powerful cloud platform that offers you the best tools for your AI projects with responsible AI built-in.

順便說一句，Azure AI 是一個強大的雲平臺，內置負責任的人工智能，可為您的人工智能項目提供最好的工具。

And here comes the best part, you can even try it out for free through the link in the video description.

最精彩的部分來了，你甚至可以通過視頻描述中的鏈接免費試用。