什么叫真正的元宇宙级别虚拟数字人？

人们拥抱数字人是因为相信数字“人”在上面两个超越真人价值的同时，能够像真人一样满足各种场景。那就要求数字人能够出席参加各种活动、能够实现作为人的各种职责、能够像人一样满足我们对内容的各种想象。能达到这些要求的数字人就可以满足各种元宇宙场景，可以称之为元宇宙级数字人。

People embrace digital people because they believe in numbers & & ldquo; people & & rdquo; the two above-mentioned things that transcend the values of the real person can satisfy the scenes as if they were real. So they are required to be able to attend events, to be able to fulfil their responsibilities as human beings, and to be able to meet our imaginations of content as if they were human beings.

图源：网络

当你发现了一个数字人代替真人做了代言人，就会期待他代替真人参加各种活动，完成各种影视作品，各种直播，演技高超，情感真挚。

When you find out that a digital person is acting as a spokesperson for the real person, you expect him to take the place of the real person in a variety of events, in a variety of visual works, in a variety of live broadcasts, in a highly skilled and emotional way.

反之，如果数字人做不到这一点，就像自动驾驶一样，前几年宣传的时候大家都以为很快就不用自己开车了，可多年过去，还遥遥无期，数字人如果也是这样，就会是空中楼阁！

On the other hand, if digital people can't do this, like autopilots, they thought they wouldn't be driving by themselves in the past few years, but they'd be gone for years, and if they were, they'd be in the air!

所以除了数字人本身的新价值之外，根据上面多个元宇宙级细分场景的要求，并通过分析真人推导出数字人的能力后, 小编将能满足这些场景、拥有这些能力的数字人称之为——元宇宙级数字人，并从技术上总结了三个要素：

图源：网络  

• 高级别:

静态达到影视界别,PBR效果逼真、动态精准传达真人的喜怒哀厌、说话口型、动作精准、自由多样。有灵魂、有情感、有理解能力。

Stereotypes reach the visual divide, the PBR effect is real, dynamically and accurately conveys the joy of the real person, speech, precision, freedom and diversity. There are souls, emotions, and understanding.

• 实时驱动:

数字人实时驱动。可以在元宇宙中呈现出活灵活现的人。

Digital people are driven in real time. People can be present in the meta-cosm.

• 普及易用:

单个数字人成本万元级。普通摄像头、普通衣服。轻松易用、人人可用。

A single digital person costs tens of thousands of degrees.

以上元宇宙级数字人三要素，至少能满足前两要素，只是规模小一点，还可以称之为“小元宇宙”,否则就容易误导大众做不切实际的想象。

The three elements of the meta-cosm level digital person above, which satisfy at least the first two elements, are small enough to be called & ldquao; the mini-cosm and & rdquao; otherwise, it is easy to mislead the public into making unrealistic assumptions.

小编从所有数字人中搜集了一些有代表性的，快来看看是否达到了你心中元宇宙级的标准？

Editor-in-chief collects some representative data from all the digital people to see if the meta-cosm level of your mind has been met.

一、图片视频类：

i. Photo video class:

图源：网络

据柳夜熙官方宣传她们的动画制作成本1/秒=1克黄金的价格。这些你以为的虚拟人其实都是由传统手key动画流程制作的,以图片、视频方式展现，跟10年前20年前动画片的人物本质上没任何区别。只能说是动画效果做的还不错，但很遗憾：粉丝看不到她们的直播也没办法和他们实时互动，所以这些不能纳入元宇宙级别,

According to Willow's official campaign, the cost of their animated production is one second = one gram of gold. These virtuals, as you think, are made by traditional hand-key animated processes, displaying them in pictures and videos, are essentially no different from the characters of the animated films 10 years ago 20 years ago. The animated effects are good, but unfortunately, fans cannot interact with them in real time, so they cannot be included in the meta-cosm level.

二、对话/自动播报机器人类：

II. Dialogue/auto-broadcast human:

图源：网络

这些有的看着还很像真人，骗到你了吗？其实她们就是用真人的图片、视频进行换脸、换嘴合成的像素人。

Some of these look like real people. Did they lie to you? They're actually people who use real people's pictures and videos to change their faces and to synthesize their mouths.

不管是像素生成还是模型制作的CG动画,都不能在原来视频的基础上换角度、换光照、换发型、换服装、换动作。不能像真人一样根据要讲述的内容自由自在的做动作，表达情绪，更不能和观众进行像真人一样的互动，没有离魂、没有情感、没有理解。只能在播报新闻资讯、查询特定信息时比较方便。也受到众多网友评论调侃: “你以为的人工智能其实是人工智障, 只能做个工具人”。

CG animations, whether made of pixels or models, cannot change angles, lights, hairs, costumes, moves on the basis of the original video. You can't express emotions like the real person, or interact like the real person, without feeling or understanding. It can only be easier to broadcast news and search for specific information. You can also get a lot of comments from online users:   & & ldquo; you think artificial intelligence is actually artificial,   you can only be a tool & & rdquao;

三、低级别类（只能玩玩用的）：

3 low-level category (play only):

图源：网络

以上这些只能玩玩的低级别类共性就是表情弱、没有感染力、只有上半身简单的肢体动作。

These low-level homogeneity categories, which can only play, are weak expressions, non-infective, and simple physical movements in the upper part of the body.

往往我们中之人都感动哭了，这些卡通形象看起来还在卖萌，严重影响观感。很多时候我们看这些数字人直播表达的情绪完全靠听声音猜，简直就是在看音频主播。现在B站上已经有几万这样的主播了，很多都没人看。大家也都说新做这样的主播不可能再火了，原因就是达不到真人的表现力，所以没有基石价值，真人主播几百万都照样可以继续火，就是因为表情动作准确，感染力强。这和小编说的数字人第一要素：高级别的要求都差远了!也不能属于元宇宙级。

Often, we all cry, and these cartoon images look cute and seriously affect the perception. Most of the time, we watch these numbers live, and we listen to sound guesses, and it's like watching audio anchors. There are already tens of thousands of such anchors on station B. Many of them are missing. Everyone says that new anchors like this can no longer be hot, because they don't have the power to act, so they have no foundational value, and millions of them can continue to fire, just because their faces are accurate and contagious.

真正可以称之为元宇宙级的数字人终于来啦!

图源：网络 

终于等到了能满足一个个元宇宙级场景的数字人了，此类的数字人驱动技术实现了传统影视行业导演预览的效果、可以精准捕捉人的无穷表情跟全身肢体动作、并能实时驱动直播落地活动!

Finally, we have a digital figure who can satisfy a meta-cosm level scenario. Such digital driver-driven technology has achieved the effects of traditional video industry directors' previews, is able to capture people's endless expressions and body movements, and is able to drive live live landings in real time!

 代表公司: Vicon、optitrack、青瞳、xsense、dynamicxyz、cubic motion（图源：网络）

这类是在影视行业里用了很多年的传统专业动捕面捕设备，可以达到，但穿着与定制繁琐，价格昂贵。他们的共同点是:有场地限制需要在独立半封闭空间，每次表情捕捉都需要定制面部贴点、全身穿戴动捕服装、提前准备1个小时的匹配和校准、使用者需要1周以上专业培训、10人以上团队的支持维护;换人成本高、每次换人就要重新定制、培训、熟悉操作等1个月以上的时间;且直播过程中主播不能休息，设备穿戴一次就需要重新适配，大大降低了使用效率。

Such devices, which have been used for many years in the visual industry, can be achieved, but are cumbersome and custom-made and expensive. Their common features are: space limitations require an independent semi-enclosed space, each face capture requires a custom-made face spot, a full body-dressing suit, an advance one hour of matchmaking and calibration, user needs more than one week of professional training, support maintenance by teams of 10 or more, high replacement costs, re-customization, training and familiarization for more than one month, etc.; and the main broadcast during the live broadcast cannot rest and the equipment needs to be re-assembled once, significantly reducing the efficiency of use.

再一看价格，一套驱动设备下来大几十万至上百万，抱歉小编家境贫寒可能不配，先行告辞了～难道我们个人主播、小公司就没有机会踏足元宇宙了吗？不急不急…

If we look at the price, we'll have hundreds of thousands of millions of dollars in a set of drivers, and we're sorry that the small editor's poor family may not be right.

普通摄像头实现虚拟数字人拍戏，虚拟老婆演技打几分？

笔者最近在B站发现了一个 “ 赛博演猿cybactor ” 的技术演示的视频，仅用一个普通家用摄像头就能实现面部微表情、全身肢体动作的精准捕捉，来自于聚力维度。让人眼前一亮不禁畅想如果未来都能够普及这种应用效果的驱动设备，那元宇宙的第一批驻民远超10亿不是梦，人人都是头号玩家了。现在疫情再次爆发，小编正在居家隔离，期待跟大家共同实现元宇宙到来的那天都能拥有帅气漂亮的元宇宙级虚拟数字人。

I recently discovered a & ldquo; Sabo plays ape cybactor & rdquo at Station B; and the technology demonstration video, using a simple home-based camera alone, captures facial micro-expressions, body-to-body movements, and comes from a concentration dimension. It makes people wonder that the first inhabitants of the meta-cosmos are far more than 1 billion people than they dream, and everyone is the number one player. Now the epidemic is happening again, and the editor-in-law is living in isolation, looking forward to having a beautiful meta-cosm digital figure with you on the occasion of the realization of the meta-cosmos.

*声明：企业通稿非VR陀螺官方稿，法律问题一律与VR陀螺无关。