标题: 【TED演讲】捕捉人物面部表情（文字版） [打印本页] 作者: 半岛的铁盒 时间: 标题: 【TED演讲】捕捉人物面部表情（文字版） One of the biggest challenges in computer graphics has been being able to create a photo-real, digital human face. And one of the reasons it is so difficult is that, unlike aliens and dinosaurs, we look at human faces every day. They are very important to how we communicate with each other. As a result, we're tuned in to the subtlest things that could possibly be wrong with a computer rendering, in order to believe whether these things are realistic. 计算机图形制作面临的巨大挑战之一是如何制作一个真实的人物面部表情。不像外星人和恐龙，它之所以如此困难的原因是我们每天都会看到人的面部表情面部表情对我们非常重要，因为我们通过表情来彼此沟通由此可见我们会注意那些非常微妙的东西通过计算机直接渲染出来的很可能是不正确的更无法使人相信这些是真实的
And what I'm going to do in the next five minutes is I'm going to take you through a process where we tried to create a reasonably photo-realistic computer-generated face, using some computer graphics technology we've developed, and also some collaborators at a company called Image Metrics. And we're going to try to do a photo-real face of an actress named Emily O'Brien, who is right there. And that's actually a completely computer-generated rendering of her face. By the end of the talk, we're going to see it move. 在接下来的5分钟里面我将带领大家一起浏览一个流程来通过计算机创造一个写实的人物表情通过我们开发的一系列计算机图形技术同时和一些来自Image Metrics公司的伙伴我们将尝试制作一个写实的名叫艾米丽(Emily O'Brien)的女演员的表情，她就在那儿。那实际上是一个完全由计算机渲染生成的她的面部头像。在讲座的结尾，我们将看到她动起来
The way that we did this is we tried to start with Emily herself, who was gracious enough to come to our laboratory in Marina Del Rey, and sit for a session in Light Stage 5. This is a face-scanning sphere, with 156 white LEDs all around, that allow us to photograph her in a series of very controlled illumination conditions. And the lighting that we use these days looks something like this. We shoot all of these photographs in about three seconds. 我们尝试的方式是通过Emily她自己她非常优雅高尚地进入我们的实验室在玛丽安德尔湾，并坐到了我们的设备灯光舞台5当中这是一个面部扫描球体，有156个白色的发光二极管分部在其中，配合我们对她进行拍照通过一系列严格控制的光照条件我们近期使用的光照方式看起来是这样的。我们在三秒钟之内拍摄出全部这些照片。
And we basically capture enough information with video projector patterns that drape over the contours of her face, and different principle directions of light from the light stage, to figure out both the coarse-scale and the fine-scale detail of her face. If we zoom in on this photograph right here, we can see it's a really nice photograph to have of her, because she is lit from absolutely everywhere at the same time to get a nice image of her facial texture. 我们主要地会采集足量的信息通过图像投影模拟出她的面部轮廓以及通过灯光舞台得到的不同角度光照条件下的数据来计算出粗略的骨骼模型以及她的精细的面部细节模型数据如果我们看放大的这张照片我们可以很清晰地看到她的面部细节因为她的面部被同时从每个角度完全地照亮了来得到一份非常高质量的她的面部表情贴图
And in addition, we've actually used polarizers on all the lights — just like polarized sunglasses can block the glare off of the road, polarizers can block the shine off of the skin, so we don't get all those specular reflections to take this map. Now, if we turn the polarizers around just a little bit, we can actually bring that specular reflection of the skin back in, and you can see she looks kind of shiny and oily at this point. If you take the difference between these two images here, you can get an image lit from the entire sphere of light of just the shine off of Emily's skin. 并且，我们实际上在所有灯上面应用了偏振镜就如同偏振的太阳镜可以阻挡马路上的耀眼光斑偏振镜可以组织皮肤上出现光芒所以我们没有采集到这些贴图的高光反射部分现在，如果我们稍微调整一下偏振镜我们就可以得到高光的反射在皮肤上面并且，你会看到她的面部看起来有些闪耀和油光。如果你取一下两张图像的差你得到的合成图像就是360度的光照条件下艾米丽(Emily)皮肤的光照效果
I don't think any photograph like this had ever been taken before we had done this. And this is very important light to capture, because this is the light that reflects off the first surface of the skin. It doesn't get underneath the translucent layers of the skin and blur out. And, as a result, it's a very good cue to the detailed shape of the skin-pore structure and all of the fine wrinkles that all of us have, the things that actually make us look like real humans. 我认为没有任何图像会这样拍摄在我们这样做之前。采集光照信息是非常重要的，因为光照会反射出皮肤表面的最上一层它不会出现在下层半透明层，它会模糊掉。由此可见，它是一个很好的线索来获得皮肤细致的毛孔结构以及我们都有的皮肤褶皱细节的全部，实际上，这些都使得它看起来更接近人类的真实效果。
So, if we use information that comes off of this specular reflection we can go from a traditional face scan that might have the gross contours of the face and the basic shape, and augment it with information that puts in all of that skin pore structure and fine wrinkles. And, even more importantly, since this is a photometric process that only takes three seconds to capture, we can shoot Emily in just part of an afternoon, in many different facial poses and facial expressions. 如果，我们使用来自高光反射的信息我们可以通过传统的面部扫描而得到面部的粗略的轮廓和基本的形状，同时输入大量的信息皮肤的毛孔结构和纹理特征更重要的是，因为这是基于照片建模的流程而我们只采集了3秒钟的照片数据我们可以用一下午的时间拍艾米丽(Emily), 拍摄不同的面部姿态和表情变化。
So, here you can see her moving her eyes around, moving her mouth around. And these we're actually going to use to create a photo-real digital character. If you take a look at these scans that we have of Emily, you can see that the human face does an enormous amount of amazing things as it goes into different facial expressions. You can see things. Not only the face shape changes, but all sorts of different skin buckling and skin wrinkling occurs. You can see that the skin pore structure changes enormously from stretched skin pores, to the regular skin texture. You can see the furrows in the brow and how the microstructure changes there. You can see muscles pulling down at flesh to bring her eyebrows down. Her muscles bulging in her forehead when she winces like that. 这里，你可以看到她转动眼睛，嘴巴的开合。这些，实际上都会被我们用来创造数字虚拟的写实人物。来看一下这些我们对艾米丽(Emily)进行的扫描你会看到我们人类面部表情是千变万化的在不同的面部表情变换之中你会看到，不仅仅是面部形态发生变化以及各种各样的皮肤的弧线和皮肤上的褶皱出现。你可以看到皮肤毛孔结构也有着巨大的变化从被拉伸的皮肤毛孔，到普通的皮肤质地你可以看到前额眉骨那里细微的变化。你可以看到肌肉的收缩在她皱眉的时候。她前额的肌肉会这样的膨胀，在她畏惧的时候。
In addition to this kind of high-resolution geometry, since it's all captured with cameras, we've got a great texture map to use for the face. And by looking at how the different color channels of the illumination, the red and the green and the blue, diffuse the light differently, we can come up with a way of shading the skin in the computer. Then, instead of looking like a plaster mannequin, it actually looks like it's made out of living human flesh. 对于这样高分辨率的几何体，因为这些都通过照相机采集，我们有足够的她的面部表情贴图。通过颜色和亮度不同的通道，红绿蓝色彩通道受光照不同的漫反射我们可以通过计算机总结出为皮肤添加阴影的方式。那么，不同于看起来如同石膏像的假人，她看起来如同人类一样鲜活。
And this is what we used to give to the company Image Metrics to create a rigged, digital version of Emily. We're just seeing the coarse-scale geometry here. But they basically created a digital puppet of her, where you can pull on these various strings, and it actually moves her face in ways that are completely consistent with the scans that we took. And, in addition to the coarse-scale geometry, they also used all of that detail to create a set of what are called "displacement maps" that animate as well. these are the displacement maps here. And you can see those different wrinkles actually show up as she animates. 这就是我们使用的为Image Metrics公司制作的写实的数字虚拟人物Emily。这里，我们看看一下粗糙的几何模型但他们已经对她创建了一些基本的数据程式，你可以拖拽这些不同的坐标，它实际上运动的方式就会依照我们扫描出的数据那样。并且，对于这个基本模型，他们也使用了全部的细节来创建一些列的置换贴图同样可以对其进行动画设置。这里是一些置换贴图同时你可以看到不同的褶皱出现在她的动画之中。
So the next process was then to animate her. We actually used one of her own performances to provide the source data. So, by analyzing this video with computer vision techniques, they were able to drive the facial rig with the computer-generated performance. So what you're going to see now, after this, is a completely photo-real digital face. We can turn the volume up a little bit if that's available. 所以，接下来的步骤是对她进行动画设置。我们实际上应用她自身的表演作为源数据。通过计算机视觉技术来分析这些视频，他们可以驱动表情的变化通过计算机的预置效果生成。所以，我们将看到，通过这一切处理，是一个完全写实的数字话面部。我们可以开大一点声音，如果可以的话。
Emily: Image Metrics is a markerless, performance-driven animation company. We specialize in high-quality facial animation for video games and films. Image Metrics is a markerless, performance-driven animation company. We specialize in high quality facial animation for video games and films. Emily:Image Metrics是一家追求高效的动画公司我们致力于制作高质量的面部表情动画为视频游戏和电影。 Image Metrics是一家追求高效动画的公司。我们致力于制作高质量的面部表情动画为视频游戏和电影。
Paul Debevec: So, if we break that down into layers, here's that diffuse component we saw in the first slide. Here is the specular component animating. You can see all the wrinkles happening there. And there is the underlying wireframe mesh. And that is Emily herself. 保罗·德贝维奇：如果我们把它分解到各个层，这是漫反射部分我们在第一张幻灯片里面看到的。这是高光部分的动画。你们可以看到各处皮肤褶皱的出现。这是原始的线框结构层。那是Emily她自己。
Now, where are we going with this here? We've gone a little bit beyond Light Stage 5. This is Light Stage 6. And we're looking at taking this technology and applying it to whole human bodies. This is Bruce Lawmen, one of our researchers in the group, who graciously agreed to get captured running in the Light Stage. And let's take a look at a computer-generated version of Bruce, running in a new environment. And thank you very much. (Applause)现在，我们接下来会做些什么？我们已经有了高于灯光舞台5的装置设备。这是灯光舞台6. 我们期望将这一技术应用到人的全身。这是布鲁斯劳曼，我们研发团队中的一位，非常乐意配合我们采集他在灯光舞台中跑步的数据。下面让我们看一下计算机生成的布鲁斯的影像，在一个全新的环境中跑步。非常感谢。（掌声）