瞭望元宇宙|数字人闯入元宇宙 虚拟形象究竟蕴含多少可能?

资讯 2024-06-21 阅读:75 评论:0
美化布局示例

欧易(OKX)最新版本

【遇到注册下载问题请加文章最下面的客服微信】永久享受返佣20%手续费!

APP下载   全球官网 大陆官网

币安(Binance)最新版本

币安交易所app【遇到注册下载问题请加文章最下面的客服微信】永久享受返佣20%手续费!

APP下载   官网地址

火币HTX最新版本

火币老牌交易所【遇到注册下载问题请加文章最下面的客服微信】永久享受返佣20%手续费!

APP下载   官网地址

  元宇宙(Metaverse),这个来源于科幻小说的概念,已成为真实世界中的流行语。围绕这一新兴概念,一场产、学、研的实践正在展开。数字化转型中,元宇宙能否担当大任?这些新概念在中国语境下如何落地?南财合规科技研究院数字娱乐课题组策划了《瞭望元宇宙》系列报道,以期为行业和社会公众提供理解元宇宙的敲门砖。

In the context of this emerging concept, a production, learning, and research exercise is under way. In the context of the digital transition, can the Yuan cosmos take up a role? How do these new concepts land in the Chinese language?

  数字人作为元宇宙的“原住民”,成为了近年来业界与学界讨论的重点,目前业界普遍认为,数字人的大规模应用依赖于高写实度的效果、好的生产效率和可控的成本。但与此同时,终端的运行叠加算力上的需求对实时云渲染能力形成了挑战。系列报道第六篇,我们探讨元宇宙赖以运行的底层技术,以及其背后的实际产业应用。

As a “indigenous” of the meta-cosm, the Digital Man has been the focus of discussions in industry and academia in recent years, and it is now widely accepted in industry that large-scale applications of digital people are dependent on high-text reality, good production efficiency, and manageable costs. At the same time, however, the need for real-time cloud resonance is challenged by the value of the terminal's running super-calculation.

  近年来,随着人工智能虚拟现实等新技术的快速发展,虚拟数字人开始进入普通人的生活。而在新奇的外观形象、灵巧的互动效果之外,行业也正在挖掘出数字人更多的潜力。

In recent years, with

  “未来数字人的发展,正呈现出静态向动态、2D向3D、卡通向写实、单向向多向互动的趋势。”蔚领时代创始人兼CEO郭建君在接受21世纪经济报道专访时表示,数字人领域目前已经度过萌芽期,进入了快速成长阶段,不久的将来会迎来爆发期。这也意味着行业开始进入商业化变现探索的关键阶段,标准化也将逐渐建立。

“The future of digital human development is showing a static dynamic, 2D to 3D, cartoons to be written, and one-way multi-way interaction.” In an interview with the 21st Century Economic Report, the founder of the era and CEO Quixiang said that the digital sphere is now emerging and is entering a rapid growth phase, with an outbreak in the near future.

  但在丰富的想象空间之外,一些落地的问题仍然困扰着从业者与用户。在未来,数字人是否可以实现高度实时性,高度可灵活性和高度适应落地场景,并与人类进行有效的交互,仍有待时间去验证。

But beyond the wealth of imagination, there are problems of landings that still haunt practitioners and users. In the future, it is still time to verify whether digitals can achieve a high degree of real-time, high degree of flexibility and adaptability to landing scenarios and to interact effectively with humans.

  面临品质成本双难题

Quality cost dilemma.

  中国人工智能产业发展联盟发布的《虚拟数字人发展白皮书》显示,“虚拟数字人”一词最早源于1989 年美国国立医学图书馆发起的“可视人计划”(Visible Human Project, YHP)。2001年, 国内以“中国数字化虚拟人体的科技问题”为主题的香山科学会议滴174次学术讨论会提出了“数字化虚拟人体”的概念。

China

  但在当时,这些“虚拟数字人”主要是指人体结构的可视化,以三维形式显示人体解剖结构的大小、形状、位置及器官间的相互空间关系。与当下所探讨的元宇宙中的“原住民”数字人有较大的含义出入。

At the time, however, these “virtual digital figures” referred mainly to visualization of the structure of the human body, showing the spatial relationship between the size, shape, location and organs of anatomy structure in a three-dimensional form.

  近五年得益于深度学习算法的突破,数字人的制作过程得到有效简化,虚拟数字人开始步入正轨。

Over the past five years, thanks to a breakthrough in deep learning algorithms, the digital production process has been effectively streamlined and virtual digitalists are beginning to get on track.

  同时,这也吸引到政策层面的关注。

At the same time, this has attracted attention at the policy level.

  2022年8月,北京市经济和信息化局发布了国内首个数字人产业专项支持政策——《北京市促进数字人产业创新发展行动计划(2022—2025年)》(以下简称《计划》)。《计划》提出,抓住以数字人为代表的互联网3.0创新应用产业机遇,充分发挥国际科技创新中心建设优势,打造数字人产业创新高地。

In August 2022, the Beijing Municipal Bureau for Economics and Informatization issued the first specific support policy for the digital industry in the country - the Beijing Municipal Action Plan for Innovation in the Digital Industry (2022-2025) (hereinafter " the Plan " ). The Plan proposes to seize the digitally represented opportunity of the Internet's 3.0 innovation application industry, to take full advantage of the International Centre for Science, Technology and Innovation in order to create a high ground for innovation in the digital industry.

  《计划》提出了发展目标,“到2025年,北京将培育1—2家营收超50亿元的头部数字人企业、10家营收超10亿元的重点数字人企业,建成10家校企共建实验室和企业技术创新中心,打造5家以上共性技术平台,培育20个数字人应用标杆项目,建成2家以上特色数字人园区和基地等目标。”

The Plan sets out development goals: “by 2025, Beijing will have cultivated one to two leading digital enterprises with more than RMB 5 billion in revenue, 10 leading digital enterprises with more than RMB 1 billion in revenue, 10 joint laboratories and enterprise technology innovation centres in 10 schools, more than five common technology platforms, 20 digital application poles, and more than two unique digital parks and bases.”

  但行业现状似乎与《计划》提出的目标仍有一定距离,21世纪经济报道记者多方了解到,目前来看,一些企业在打造数字人的过程中仍会遇到渲染品质较低、制作成本高昂,IP打造周期长等难题。

But the industry seems to be a bit off track with the goals set out in the Plan, and the 21st Century economic reporters have learned that

  京东集团副总裁、京东智能产品与服务部总裁何晓冬在接受21世纪经济报道采访时表示,数字人是数字原生、人机交互等技术落地过程中最被寄予希望的形态之一,但在实际应用的过程中,有些数字人的动作十分僵化,只能起到展示的效果,无法提供实际办理业务的交互。

  何晓东以银行场景的数字人举例称,如果数字人要把整套银行业务办理好,需要克服几个非常落地的问题。“首先是需要避免噪音,因为银行柜台是比较噪杂的场所。其次是银行数字人需要和后台的业务支持系统有效地对接,必须能够解决很多银行的具体业务问题。”何晓东举例称,转账、取钱、改密码等业务行为都意味着数字人在知识推理、语音推理等方面都要达到很高标准,因此这些方面的同步也是很大的挑战。

  三种商业模式

Three business models.

  郭建君在采访中也表达了类似看法,他表示,数字人成熟的背后一方面是技术整合能力,另一方面是IP运营能力。“数字人是一个产业链很长的行业,既需要底层引擎技术的发展,也需要对上层工具和生产管线的优化,才能同时满足数字人形象好、生产效率高和成本可控的三重效果。”

Kwok Jianjun expressed a similar view in his interview, stating that the skills behind digital maturity are, on the one hand, technological integration and, on the other hand, IP operations. “Digital people are a very long industry chain that requires both the development of bottom-level engine technology and the optimization of upper-level tools and production tubes in order to satisfy the triple effect of good digital image, productivity efficiency and cost-controllability.”

  而IP运营能力体现在对数字形象打造的思路和经验之上。郭建君表示,当前行业较为普遍的情况是,技术制作更有信心,运营思维比较欠缺。在这方面,蔚领时代正在探索的一条道路是——选择更专业的生态伙伴合作进行商业化。

According to Kwok Jianjun, the current industry is more confident in technology production than in business thinking. In this regard, one of the paths being explored in the era of assertiveness is the choice of a more professional ecological partner for commercialization.

  近日,蔚领时代与文娱企业海西传媒集团组建了独立运营的公司蔚海灿娱,推出了龚俊的数字人“霁风”,郭建君表示,后续将围绕可互动数字IP的矩阵搭建、共创虚拟空间的构建等方面进行多重合作。

In recent days, the Heixi Media Group, a leading and entertainment company, has set up an independent company and launched Gon Jun's Digital Man, and Guo Jianjun has indicated that there will be multiple collaboration around the construction of an interactive digital IP matrix and the creation of a virtual space.

  值得一提的是,利用契合特定场景的IP加快表演型数字人创新也是《计划》的规划的主要任务之一,其提出要“促进技术人才与艺术人才联合创新,加快数字人IP孵化,培育创作者经济。同时打造精品数字人品牌,支持数字人参与广告营销、品牌代言,加强数字人运营管理,树立传播正能量的数字人形象。”

It is worth mentioning that one of the main tasks of planning for the Plan is to accelerate digital innovation using IPs that match specific scenarios, and it proposes “to promote innovation that combines skilled and artistic talent, accelerates the incubation of digital IPs, and fosters the creative economy.” At the same time, it is important to create digital brands that support digital participation in advertising marketing, branding, strengthening the management of digital operations and creating digital images that transmit positive energy.”

  事实上,IP运营,客户服务和流量变现正是当前数字人主要的三种商业模式。

In fact, IP operates, and client services and traffic changes are now the three main business models for the current digital population.

  腾讯云近日发布的《数字人产业发展趋势报告(2023)》指出,内容/IP型数字人主要依靠数字人的商业IP运营实现商业变现,功能服务型是交付给客户实现变现,而虚拟分身则需要结合虚拟空间的设定和规则进行流量变现。

According to the Report on Trends in the Development of the Digital Human Industry (2023), recently released by Xeng Xinjung, the content/IP-type digital person relies mainly on the commercial IP operation of the digital person for commercial realization, the functional service type is delivered to the client for realization, while the virtual dichotomy requires flow realization in the context of virtual space creation and rules.

  “数字人的制作成本和周期伴随定制化程度的下降而递减。三种数字人的制作方式和周期也存在差异,内容/IP型数字人更多是项目制,以定制为主;功能服务型是产品和定制化相结合,在通用产品上,进行定制化开发;而虚拟分身更多是产品制,用户利用平台工具产品进行数字人形象的制作并在相应虚拟场景内进行活动,一般而言,内容/IP型数字人的成本和周期最高,虚拟分身型数字人最低。”

“Digital production costs and cycles are reduced by a reduction in customization. There are also differences in the production patterns and cycles of the three digitals, with content/IP digitals more project-based and custom-based; functional service types are combinations of products and customization, custom-based development on generic products; and virtual compartmentalization is more product-based, with users using platform tool products for digital image production and activities within the corresponding virtual setting, with content/IP digitals generally having the highest costs and cycles, and virtual split digitals having the lowest.”

  大规模应用依赖可控成本

Large-scale application dependent on manageable costs

  “数字人的大规模应用,有三个条件必不可少:高写实度的效果、较高的生产效率和可控的成本。”在郭建君看来,虽然对比传统离线渲染的影视级别的写实度效果仍然需要提升,但当前行业应用的实时渲染的技术,在生产效率、可控成本和数字人可互动性上均具备了优势。

“The large-scale application of digital people is essential for three conditions: high-writing effectiveness, high production efficiency and manageable costs.”

  科技部信息通信专家、上海流程智造科技创新研究院院长贺仁龙告诉21世纪经济报道记者,构造虚拟世界的技术正在逐渐成熟,应用的场景也日益丰富,效果也十分逼真。但在感知交互方面,数字人仍然面临着一系列问题。

The Information and Communication Specialist of the Ministry of Science and Technology and the President of the Shanghai Institute for Flows, Science, Technology, and Innovation Research told economic reporters in the twenty-first century that the technology to construct a virtual world was maturing, and that applications were becoming increasingly rich and effective. But digital people still faced a number of problems in understanding interactions.

  “在创建元宇宙的过程中,会构造一个世界的秩序和逻辑,但这个世界是否有重力,重力又是多少,目前来看都是技术需要探索的方向。同时,虚拟世界的逻辑需要规律和算法生成,在这其中,如何有效地使数字人在视觉、听觉之外,与用户进行触觉感知交互,也是行业面临的难题。”

“In the process of creating the meta-cosmos, the order and logic of a world will be constructed, but whether there is gravity in the world and how much gravity it is, is the direction that technology now needs to explore. At the same time, the logic of the virtual world requires the production of patterns and algorithms, in which it is also a challenge for the industry to effectively interact with the tactile senses of the users, outside the sight and hearing.”

  “未来超写实的世界将以可视化为主要特征。而这需要海量的图形运算。”郭建君认为,VR/AR终端将成为元宇宙的入口,但终端的运行依赖算力高速的运行,这其中,实时云渲染能力解决的就是把超写实的世界带到终端体验上的问题。

"The future super-writing of the world will be a major feature of visualization. This will require graphic calculations of the magnitude. Kwok Jianjunk believes that the VR/AR terminal will be the gateway to the metacosystem, but that the terminal's operation depends on the high speed of arithmetic, in which real-time cloud rendering solves the problem of bringing the overwritten world to the end experience.

  “只有云端实时渲染在达到较高标准的情况下,才能真正带来足够写实、互动、有沉浸感的图像画质。这也就是蔚领时代认为的GaaS(Graphics as a Service)时代到来的逻辑依据。”

"Only real-time cloud rendering, if it meets a high standard, can it actually produce enough graphics to be written, interactive, and immersed. That is the logic behind the age of Gaaphys as a Service, as it was considered in the Axis era."

  郭建君认为,未来支撑实时云渲染的底层技术主要是云原生的技术,比如多GPU协同渲染,以及C+S架构来实现的单个进程输出几十路视频流供几十个玩家游玩的技术。“这些创新的云原生的技术是未来元宇宙的支撑,同时围绕着类似云游戏的串流能力和云端虚拟化的能力,也是未来元宇宙所需要的。”

According to Kwok Jianjun, the underlying technologies that underpin real-time cloud rendering in the future are mainly cloud-based technologies, such as multiple GPU co-reflations, and the individual processes that the C+S architecture achieves to produce dozens of paths of video flow for dozens of players. "These innovative cloud-based technologies are the backbone of the future meta-codes, which are built around games like and the ability of the future meta-cosms to virtualize.

  “一个可喜的现象是我们看到头部渲染引擎的技术发展迅速,现在已经开始进行影视级效果的应用了。”郭建君说道。

“It is a welcome sign that we have seen the rapid technological development of the head-reflection engine and that the application of the visual level has now begun.” Mr. Guo Jianjun said.

  在云原生技术之外,3D重建技术的发展也为数字人的大规模落地带来了更多的可能性。

In addition to cloud-based technologies, the development of 3D reconstruction technologies offers additional possibilities for large-scale digital landings.

  上述《报告》指出,2D数字人和3D数字人生成/建模方式不同,2D数字人本质是生成图像,因此主要方式是利用深度学习技术根据视频生成与真人相同的图像,技术相对成熟。

The above-mentioned Report notes that 2D digitals and 3Ds are generated/modelled in different ways and that the 2D digitals are, by their very nature, the generation of images, thus mainly through the use of in-depth learning techniques to generate images based on video that are identical to those of the real person, with relatively mature technology.

  “3D数字人需要建立3D模型,主要是利用软件传统的手工建模、静态扫描建模(相机阵列扫描建模)、动态光场或者AI建模(3D重建技术)生成3D模型。目前手工建模和静态扫描建模技术较为成熟且应用广泛,利用3D重建技术或动态光场重建可极大节约建模时间,提升建模效率,是未来重要布局方向。”报告指出。

“3D digitals need to build 3D models, mainly using software traditions for manual modelling, static scanning modelling (camera array scanning modelling), dynamic optical fields or AI modelling (3D re-engineering technology) to generate 3D models. Currently, manual modelling and static scanning modelling techniques are mature and widely applied, using 3D reconstruction techniques or dynamic optical re-establishment can greatly save modelling time and improve modelling efficiency, which is an important future layout direction.” The report states.

  而在科大讯飞总裁吴晓如看来,机器想更自然地与人类进行交互,需要通过听觉、视觉、语义以及各种传感器的组合去获取更多的有用信息,AI感知方式必然要从单模态发展到多模态,逐步拟人化。

On the other hand, at , , > >, the machine needs to interact more naturally with humans by hearing, visual, synonyms and various , >,

  2021年10月,科大讯飞正式发布AI虚拟人交互平台,它具有多模感知、多维表达、情感贯穿、自主定义四大交互特点。可以在客户的不同业务场景下进行AI虚拟人赋能,提供一站式的AI虚拟人应用服务。

In October 2021,

  “科大讯飞推出的数字人目前已经开始把视觉、手势、视线和语音结合起来,实现数字虚拟人交互的多维表达。”吴晓如在接受21世纪经济报道等媒体采访时说道。

"

(文章来源:21世纪经济报道)

(原标题:瞭望元宇宙|数字人闯入元宇宙 ,虚拟形象究竟蕴含多少可能?)

(责任编辑:91)

美化布局示例

欧易(OKX)最新版本

【遇到注册下载问题请加文章最下面的客服微信】永久享受返佣20%手续费!

APP下载   全球官网 大陆官网

币安(Binance)最新版本

币安交易所app【遇到注册下载问题请加文章最下面的客服微信】永久享受返佣20%手续费!

APP下载   官网地址

火币HTX最新版本

火币老牌交易所【遇到注册下载问题请加文章最下面的客服微信】永久享受返佣20%手续费!

APP下载   官网地址
文字格式和图片示例

注册有任何问题请添加 微信:MVIP619 拉你进入群

弹窗与图片大小一致 文章转载注明

分享:

扫一扫在手机阅读、分享本文

发表评论
平台列表
美化布局示例

欧易(OKX)

  全球官网 大陆官网

币安(Binance)

  官网

火币(HTX)

  官网

Gate.io

  官网

Bitget

  官网

deepcoin

  官网
热门文章
  • 0.00003374个比特币等于多少人民币/美金

    0.00003374个比特币等于多少人民币/美金
    0.00003374比特币等于多少人民币?根据比特币对人民币的最新汇率,0.00003374比特币等于2.2826 1222美元/16.5261124728人民币。比特币(BTC)美元(USDT)人民币(CNY)0.00003374克洛克-0/22216.5261124728比特币对人民币的最新汇率为:489807.72 CNY(1比特币=489807.72人民币)(1美元=7.24人民币)(0.00003374USDT=0.0002442776 CNY)。汇率更新于2024...
  • 0.00006694个比特币等于多少人民币/美金

    0.00006694个比特币等于多少人民币/美金
    0.00006694比特币等于多少人民币?根据比特币对人民币的最新汇率,0.00006694比特币等于4.53424784美元/32.5436 16人民币。比特币(BTC)美元(USDT)人民币(CNY)0.000066944.53424784【比特币密码】32.82795436 16比特币对人民币的最新汇率为:490408.64 CNY(1比特币=490408.64人民币)(1美元=7.24人民币)(0.00006694USDT=0.0004846456 CNY)汇率更新时...
  • 0.00015693个比特币等于多少人民币/美金

    0.00015693个比特币等于多少人民币/美金
    0.000 15693比特币等于多少人民币?根据比特币对人民币的最新汇率,0.000 15693比特币等于10.6 1678529美元/76.86554996人民币。比特币(BTC)【比特币价格翻倍】美元(USDT)人民币(CNY)0.000/克洛克-0/5693【数字货币矿机】10.6 167852976.8655254996比特币对人民币的最新汇率为:489,807.72 CNY(1比特币= 489,807.72人民币)(1美元=7.24人民币)(0.00015693 U...
  • ??今日BTC和ETH行情分析以及对BICO的看法

    ??今日BTC和ETH行情分析以及对BICO的看法
    ? 如果你刚认识我那么此刻开始你的幸福? 幸运之路正式开启!? ? 历史记录皆可追溯,往期的记录依然可查,山水相逢,皆是缘!?以后也会经常分享一些看好现货给大家! ? 本周热点 ? ? ? ? ? TON 启动公...
  • 孟洪涛谈威科夫交易法

    孟洪涛谈威科夫交易法
     大咖看市 | 判断趋势的工具(一)  原创2016-04-30孟洪涛期货日报 在讨论判断工具之前,我们先说下判断趋势。趋势包括以下几个阶段:趋势的开始,趋势在运行中,以及趋势的结束。我们判断趋势就是能够找出当前市场处于趋势的哪个阶段,以便调整交易。趋势的不同阶段伴随着交易者不同的行为,起始阶段是进场时机,结束阶段是出场时机,同时也是准备反转的进场时机。但是在交易中遇到的最头疼的问题是以上几个趋势阶段并不会白纸黑字表现出来。 ...
标签列表