张晟：用代码来解构设计

个人介绍

SELF INTRODUCTION

我叫张晟，东华大学工业设计研二，对艺术/科技/商业均有着浓厚兴趣。虽然专业是工业设计，但从不局限于工业设计。目前主要研究人工智能在设计中的应用，并已开发了几款基于人工智能的设计辅助工具。很荣幸成为这次活动的第8组的一员，主题是“代码：解码：创新”

I am Zhang Sheng, a postgraduate student from Donghua University. Although my specialty is industrial design, it is never limited to industrial design. At present, it mainly studies the application of artificial intelligence in design, and has developed several design aids based on artificial intelligence.I am honored to be a member of the 8th group of this event. The theme is "Code: Decoding: Innovation."

个人作品

MY WORKS

1

汽车手绘线稿自动上色器

Car sketch autorender

受漫画上色网站PaintsChainer的启发，我意识到人工智能在设计界可发挥巨大作用，于是结合自己以前从事汽车设计的经历，制作了一个汽车线稿自动上色工具。以下是效果截图，左边是输入的线稿，右边是自动上色结果。可访问http://privatepainter.com/ 在线玩耍。

Inspired by the comic painter website PaintsChainer, I realized that artificial intelligence can play a huge role in the design community. Therefore, in conjunction with my previous experience in car design, I created a car sketch autorender tool. The following is a screenshot of the effect. The left side is the input linework, and the right side is the automatic coloring result. Visit http://privatepainter.com/ to try it out.

2

服装设计方案生成器

AI Fashion Designer

开发汽车线稿自动上色时，我发现汽车设计领域并非应用人工智能图像生成技术的最佳领域，经过调查分析，选择服装设计领域进行进一步开发。搜集当季所有的服装发布会最新服装图片，利用人工智能技术提取出其中的设计风格，建立了一套自动给线稿着色的算法，可以在几秒内自动生成任意数量的搭配方案。帮助服装设计师更快更好的抓住潮流趋势。（由于没钱租GPU服务器，程序暂时未上线）

When developing the auto-line painting of automotive lines, I found that the automotive design field is not the best field of application of artificial intelligence image generation technology. After investigation and analysis, we choose the garment design field for further development. Collect all the latest costume images of the current season's apparel release conferences, use artificial intelligence technology to extract the design styles, and set up an algorithm for automatically coloring linear scripts to automatically generate any number of collocation schemes in seconds. Help costume designers to catch trends faster and better.(Since there is no money to rent a GPU server, the program is temporarily offline.)

3

利用图像识别来玩火影忍者结印法术

NARUTO game

某日突发奇想，用卷积神经网络来识别《火影忍者》中的结印手势来玩游戏好像很有意思，于是写了十分“中二”的程序，根据识别结果，加上忍术音效和忍术特效。源代码已上传至我的github：https://github.com/Arthurzhangsheng/NARUTO_game

以下是实时效果演示：

One day, whimsy, using image AI recognition technology to identify the "Ninja gesture" in the 《NARUTO》 to play the game seems very interesting, so I wrote a very "中二" program. You can find my source code at： https://github.com/Arthurzhangsheng/NARUTO_game

4

非编程类作品

other work

还有许多非编程类作品，包括工业设计、平面设计、摄影、服务设计等，不作详细展开，部分已整理发布在：https://www.zcool.com.cn/u/14816863

There are many non-programming works, including industrial design, graphic design, photography, service design, etc., which are not carried out in detail. Some of them have been sorted out and published in: https://www.zcool.com.cn/u/14816863

对数字化领域的看法

The view of the digital field

我选择进入数字化领域的原因很功利，因为我感到传统设计领域已经是一片红海，我们这一代设计师已经错过了传统设计发展的红利期，很难赚大钱，所以我选择“人工智能+”这一蓝海，利用新技术来实现财务自由。

设计和艺术之前被人们认为是人工智能最不可能胜任的工作之一，但随着风格迁移、GAN、阿里鹿班，arkie等一系列技术和应用出现后，理性的代码和感性的艺术设计不再泾渭分明。当前的人工智能最擅长从大数据中学习到隐含结构，而我们做设计时不也是从大量信息资料中获取设计素材吗？有人会指责说用人工智能创作，只不过是特征的重组，算不上创作，但我们人类做设计又何尝不是特征重组呢？天下设计一大抄，一流设计师抄自然界，二流设计师抄跨界，三流设计师抄同行，从来就没有真正意义上的原创。

我相信未来一定会出现AI设计大师，不过目前阶段，AI还只是个设计实习生，无法替代人类设计师，但它能帮助我们做很多简单重复的工作，提升我们的效率，这也是我目前所重点研究开发的领域之一。

The reason why I chose to enter the digital domain is very utilitarian because I feel that the traditional design field is already a Red Sea. Our generation of designers has already missed the dividend period of traditional design development and it is difficult to make big money, so I chose “artificial intelligence plus” ，a blue ocean, to achieve financial freedom with new technologies.

Design and art were previously considered to be one of the most impossible tasks for artificial intelligence, but with the appearance of a series of technologies and applications such as style migration, GAN, Ali Luban, and arkie, rational codes and perceptual artistic designs are not Again and again. The current artificial intelligence is best at learning implicit structures from big data, and we do not design materials from large amounts of information at the time of designing. Some people will accuse people of using artificial intelligence to create. It's just a reorganization of the characteristics. It's not creation, but isn't our human design a feature reorganization? The design of the world is a big copy. The first-rate designers copy the nature, the second-rate designers copy the cross-border, and the third-rate designers copy the counterparts. There has never been a true original.

I believe there will be AI design masters in the future, but at this stage, AI is only a design intern and cannot replace human designers. But it can help us to do a lot of simple and repetitive tasks and improve our efficiency. This is one of what I currently focus on.

对活动的期待

Expectation of activity

希望能认识结交各路大神，跨界合作，擦出奇妙的火花。如果能找到创业合伙人就再好不过了。

We hope to know each other and make cross-border cooperation and wipe out wonderful sparks. It would be great if I could find business partners.

最后打个小广告，欢迎各行各业人士跨界合作搞事

个人公众号/Public Accounts：

人工智能与设计

github：

https://github.com/Arthurzhangsheng

微信/wechat：

18817827974