The state of clothing manufacturing

Every industry has a holy grail – the ultimate problem that an industry needs to solve. It should be supported or driven by morality and curiosity. For example, quantum computing’s holy grail is to reach millions of qubits to simulate nature. For fashion, it is to reach inventory = 0, minimum order quantity MoQ = 1 and SKU = infinity, to reduce waste and pollution while supporting creativity. Why does apparel production need to change and innovate? It is one of the most polluting industries in the world since the first industrial revolution enabled mass production. Meanwhile, human labor has been exploited to the extreme. The industry calls its evolution “chasing the needle”, meaning that production is moved from countries to countries with lower and lower labor costs. This is unsustainable. The pandemic further exposed the problems in the supply chain. In addition, mass production reduces creativity and does not support experimentation and craftsmanship diversity of independent designers.

Here is a video I recorded last year that summarizes the problems of garment production and the direction I am exploring by establishing an independent designer brand (Art by Physicist). If you are interested in the details of the products mentioned in it, you can find them on the official website www.kittyyeung.com

To a layman, “fashion” and “technology” may seem to be very separate industries. Yet, if one looks into the process of constructing a piece of garment, they will quickly realize it is nothing but engineering. How do you cut, fold, and stitch a piece of fabric to drape nicely on a person’s body you are designing for? Would you do it by hand or get assistance from a machine?

Industry 4.0 for Fashion

Last summer, I went to Texprocess/Techtextil/Heimtextil at Messe Frankfurt GmbH and observed all the innovation areas happening in textile goods production machinery. It was the largest tradeshow I’ve ever been to, with miles and miles of exhibitors from all over the world displaying a survival scene from the pandemic and even thriving again on textile goods production. Here I summarize the types of innovation that I saw. If you were there and saw additional areas, feel free to comment below.

1. Digital printing on fabric is a mature technology. Many websites already let designers submit images to print on demand.

2. Thread printing on-demand is a new area of development, Twine Solutions It is in POC stage for sample making and future low-quantity production. Each item can be designed with a unique thread color combination without the need to switch threads for a different color. This will save factories switchover time when the quantity of orders is small.

3. Lots of options for automated printing and cutting single-ply/multi-ply which come with marker software Eastman Machine Company, Hasler, Die Zutrittsexperten / Les Experts d’accès , bullmer GmbH, Allcomp Polska, MorganTecnica, Audaces.
My favorite is expert systemtechnik gmbh’s camera detection above the cutting bed to detect the shapes and defects on a cow skin, one piece at a time, because every cow is different! The software automatically configures the patterns to avoid the defects and maximizes the layout to fill up the good leather area.

4. Machine embroidery is mature with wide adoption in factories and is an element for customization. New development is around easily configurable software, factory management system IoTs and conductive thread embroidery e.g. ZSK Stickmaschinen .

5. Sewing is considered the most difficult process to fully automate. But it has been made easier by designing machines specific to sewing certain flat elements. For example, pre-cut fabrics for pockets, collars and zippers can be sewn automatically by inserting them under the needles, sometimes compressed with plastic templates. E.g. Juki, New-tecH, PFAFF, MAICA S.R.L. MANIFATTURA ITALIANA CAMICIEJack Sewing Machine Co., Ltd.Global International B.v.

6. Sewing 3D structures are harder but there are machines programmed to sew sleeves onto armholes specifically. Dürkopp Adler

7. Some companies develop mechanical parts attached to the sewing machines to make it easier to sew along fabric curvatures Astaş Juki. And semi-automated machines only needing someone to insert the fabrics in the right slot, e.g. sleeve hems YAMATO SEWING MACHINE MFG.CO.,LTD. Juki, pleats @PFAFF

8. Robotic arms may help with pick-and-place and rotations on flat sewing beds. It is limited to sewing certain types of lines on a garment, not a full garment. Juki, KUKA, bullmer GmbH, Dürkopp Adler

9. Due to the flexible nature of fabrics, they are still difficult for robots to sew. But certain products’ sewing in 3D could be fully automated, such as the car seat sewn by KUKA in the video.

So let’s check against the industry holy grail: inventory = 0, MoQ = 1, SKU = infinity. How are the above supporting?

The industry sees reduced quantity order as a trend as consumer preference changes are faster then ever. However, specialized machines cannot meet the demand and are only efficient if they produce huge quantities of repeated units, which conflicts with the market wanting more design variations faster. Some companies approach this issue by making machines that are multi-functional.

For example, the buttonhole machine by Dürkopp Adler does various button shape sewing and cutting at once. They also made a machine with software-adjustable double needle width, one machine in the world. Companies are showing new machines as POC and gauging customer interest at the event. So can consumers drive demand of certain products, which drives retail brands to demand garment manufacturers to buy flexible machines?

Machines remain highly specialized. A new factory needs to determine what products they are going to make and only make, evaluate which machines can make these products and if it is more cost effective to hire a human to do the sewing or buy a robot replacement. Making machines easier to operate reduces the training needed to onboard workers. Modern factories will have fewer manual workers but more operation engineers to translate designs into programs for the machines. This kind of human labor is still an overhead, even if all machines are fully automated. Thus, quantity order production is still preferred over on-demand mass customization, to compensate for the labor overhead.

Therefore, for now, efficiency still means lack of creativity. If I as a creator want a new design made for Art by Physicist, I either need to get enough order first to persuade a manufacturer to make them in quantity, or I need to build my own factory with a specific set of machines to specifically make my designs or I need to limit my designs to only things these machines can make.

In order to reach the holy grail, apart from automated/semi-automated machines, it really needs each factory’s order and operation management to be set up to do mass customization, different factories coordinating based on demand of different product types, and programming software connected to design software to automatically translate how a human would construct a piece of garment to instructions the machines can replicate.

How do we prepare for industrial 4.0 starting from the small scale?

Now let’s look again at how the smallest scale can achieve on-demand production. These are now the most advanced automated print and cut machines for garments.

At Gerber Technology, a Lectra Company Innovation Center, seeing the similar Kornit Digital printer that produces my textile designs at Art by Physicist, wishing everyone were using these state-of-the-art fashion tech to make on-demand creative designs.

For Art by Physicist, I hand paint in Adobe Photoshop, overlay onto patterns in Illustrator, choose a fabric dimension at Spoonflower, upload the art there, get it printed and shipped, cut the patterns out along the outlines and sew the pieces. This way, I minimize the waste and have total control of how the artwork turns out on the outfit – a scaled down version of what’s shown in the video. Prototyping steps that can be digitized are all digitized. Steps that are still manual are due to:

1. A studio cannot afford an automated setup. Industrial printer, dryer, cutter are $…$ No lite solutions like what a sewing machine offers.

2. Could outsource to a manufacturer but development is slow (see 4 below). Require a designer to have already generated lots of sales. Nowadays I can use 3D simulation and the handmade prototype to get orders online. Fine.

3. As I design-> sell -> make, I want to fit each piece to each customer after the orders come in. That requires manual adjustment in the patterning software (I use Browzwear VStitcher).

4. Even if I take care of all above, when I give the manufacturer a teckpack to start producing, it is not turnkey. Manufacturers use different software. There are specifics to the factory a designer does not know. Both sides waste lots of time adjusting the techpack to fit the facility and redo above efforts.

5. Want to keep zero inventory and mass customization make one at a time to fit a unique customer’s bodyshape when each order comes in? 2 and 4 prohibit it. But the workflow I’m describing here is the most sustainable and inclusive.

6. White space on the fabric in the video – demo fine, in production it is all waste fabrics we try to minimize. I nest patterns in Illustrator to optimize the usage and print between patterns. The scraps are used to make small things (see collars, headbands, hats, scarfs, bags, table mats, etc. on www.kittyyeung.com) However, a manufacturer would not do this for you.

7. More efficient for manufacturers to get a batch order. Use nesting software to minimize waste and switchover. But how about nesting different styles in different customer bodyshapes?

8. Can’t just pay manufacturers to do cut-and-sew when above are not turnkey.

So how can the industry cut down waste and pollution if they can’t do the above at scale?

I’d also like to see suppliers provide more eco-friendly fabric options for digital printing and other materials with traceability.

In summary, every step from design to manufacturing should have on-demand mass customization built in with ethical sourcing, upcycling and recycling.

These are what it takes to build a fully sustainable brand and what’s keeping a PhD in Applied Physics busy on the weekend, on top of making process videos and marketing materials. If any of these intrigued you, reach out. Maybe we can generate some revenues together, to pay more people, buy some machines and donate more to STEAM education and environmental protection.

服装生产需要变革

每个行业都有一个圣杯–是一个行业需要解决的终极问题， 也是应该由道德或好奇心来支持或驱动的。比如量子计算，这个圣杯是要达到百万量子比特，才能模拟自然；而对于时尚，它是要达到库存=0，每批最小产量MoQ=1和SKU=无穷大，以减少浪费和污染同时支持创意。为什么服装生产需要变革和创新？它是从第一次工业革命开始实现了大批量生产后对世界带来最大污染的工业之一，同时劳动力被极度剥削，业内把它叫做chasing the needle， 指生产从人工费用低的国家追逐到更低的国家，这是不可持续的，一场疫情就把供应链中的问题揭露得淋漓尽致。另外，大批量生产导致创意越来越难实现，不支持独立设计师的小量尝试与多样性的工艺。

这里通过我去年录的一条视频总结一下服装生产的问题与我通过建立独立设计师品牌（Art by Physicist 物理学家的艺术）探索的方向。如果大家对里面提到的产品细节感兴趣，请到官网www.kittyyeung.com

当你看到”时尚 “和 “科技 “，是不是想到它们是毫不相干的两个行业？然而，如果我们看一下一件衣服的制作过程，会很快意识到这无非也是工程学。如何将一块布料裁剪、折叠和缝合，设计在一个人的身上的垂坠感？你会完全手工来做还是用机器来协助？

服装生产的工业4.0

上个夏天，我去了法兰克福展览中心Messe Frankfurt举办的Texprocess/Techtextil/Heimtextil，看到了最新的纺织产品生产机械中的创新方向。这是全球最大的会展中心，也是我参加过的最大的展会，来自世界各地的参展商绵延数英里，展示了从疫情中幸存下来的场景，甚至在纺织商品生产上的再次兴旺。在这里，我总结看到的创新类型，特别是关于服装方面。如果你也去了，看到了其他方面，欢迎在评论区补充。

1. 织物上的数码打印技术已经非常成熟。许多网站让设计师上传图片，按需打印，不需要大批量。

2. 按需印线的颜色是一个新的领域，比如Twine Solutions在做的。 它处于展示阶段，用于样品制作和未来的低量生产。每件商品都可以设计出独特的线色组合，而不需要在织纫过程中换不同颜色的线。这将节省工厂在订单数量少时的切换时间。

3. 有很多自动印刷和切割单层/多层的切割机选择，并且这些硬件开发商也会开发相应的排版软件，视频里展示了Eastman Machine Company, Hasler, Die Zutrittsexperten / Les Experts d’accès , bullmer GmbH, Allcomp Polska, MorganTecnica, Audaces。

我最喜欢的是Expert Systemtechnik公司在切割床上方的摄像头检测，以检测牛皮上的形状和缺陷，用软件自动根据缺陷位置排版，以避免缺陷，并最大限度地提高牛皮利用率。因为每头牛都是不同的，可以一张一张牛皮去检测再排版再切割。

4. 机器刺绣也是成熟的技术，在工厂中早已被广泛采用，也是实现定制化的一个元素。新的发展方向多在于软件方面、工厂管理系统物联网和导电线刺绣, 比如ZSK。

5. 缝纫被认为是最难实现完全自动化的工艺。但通过设计专门用于缝制某些平面元素的机器，它已经变得更加容易。例如，用于口袋、衣领和拉链的已剪好的面料可以塞到针下自动缝制，有的压在塑料模板中。 这方面的公司有很多：Juki, New-tecH, PFAFF, MAICA S.R.L. MANIFATTURA ITALIANA CAMICIE， Jack，Global International B.v.

6. 衣服的3D结构缝制还是较难，但有一些机器的程序可以专门将袖子缝到臂口上。Duerkopp Adler – Turel Technology Pvt.

7. 一些公司开发了附加在缝纫机上的机械部件，使其更容易沿着织物的弧度缝制，比如Astaş Juki。还有一些半自动的机器，只需人手把布料插入缝隙，例如缝袖子的下摆的YAMATO SEWING MACHINE MFG.CO., LTD. ，Juki, 缝褶子 PFAFF

8. 机器手臂可以帮助在布料取放和在平坦的缝制床上进行旋转运动。它仅限于在服装上缝制某些类型的线条，而不是整件服装。Juki, KUKA, bullmer GmbH, Duerkopp Adler – Turel Technology Pvt.

9. 由于织物的柔韧性，对机器手来说仍然难以缝制。但某些产品的三维缝制可以完全自动化，如视频中KUKA公司缝制的汽车座椅。

那么现在让我们来对照一下行业的圣杯，终极目标：库存=0，MoQ=1，SKU=无穷大。上述情况能如何支持呢？

业界认为，由于消费者的偏好变化比以往任何时候都快，柔性多款式小订单是一种趋势。然而，专属性的机器无法满足需求，只有在生产大量重复的单位才会有效有足够的回报，这与市场希望更快地获得更多的设计变化相冲突。

一些公司通过制造多功能的机器来解决这个问题。例如，Duerkopp Adler 的扣眼机，可以一次完成各种纽扣形状的缝制和切割。他们还制造了一台可通过软件调整双针宽度的机器，他们说这是世界上唯一一台这样的机器。是不是很惊讶？各公司在活动中展示最新机器，来衡量客户的兴趣，还没有落地应用。那么，消费者能否推动某些产品的需求，从而促使零售品牌要求服装制造商购买柔性机器呢？

机器仍然高度专属性。一家新工厂需要确定他们要做什么产品，而且只做什么产品，评估哪些机器可以做这些产品，是雇用人来做缝制工作还是购买自动化机器取代是否更划算。让机器更容易操作可以减少工人上岗所需的培训。现代工厂将有更少的手工工人，但有更多的操作工程师将设计转化为机器的程序。然而这种人力劳动仍然是一种开销，即使所有机器都是完全自动化的。所以工厂还是更希望有大订单生产而不是按需定制，以弥补劳动力的开销。

因此，就目前而言，效率仍然意味着缺乏创造力。如果我作为一个创作者想让我的品牌Art by Physicist做一款新的设计，我要么需要先获得足够的订单来说服厂商大量生产，要么我需要建立自己的工厂，用一套特定的机器来专门制造我的设计，或者我需要把我的设计限制在这些机器可以制造的东西上，因此会降低创意性。

为了达到行业圣杯，除了自动/半自动的机器，还需要每个工厂的订单管理和生产运营能做到大规模定制，不同的工厂根据不同产品类型的需求进行协调，编程软件连接到设计软件，以自动将人手如何制作一件服装的步骤转化为机器可以复制的指令。

如何从小规模开始为未来变革做好准备

那么现在再来看看最小规模如何实现按需生产吧。这些是现在最先进的服装自动化打印切割机器.

这是在Gerber Innovation Center，看到Kornit数码打印机，也就是同款给我的独立设计品牌Art by Physicist 用来打印布料图案的机器。真希望每个人都能使用这些最先进的时尚技术来做按需生产的创意设计。如果你是一个新的品牌，在无法找到愿意支持你做小批量个性化设计的厂家时可以利用更多像下面这样的数码过程，达到环保。

我在设计时用Adobe Photoshop手绘，用Illustrator叠加到版型上，在Spoonflower选择一个布料，再根据布料尺寸上传手绘，打印然后寄给我，沿着轮廓剪出版型，最后缝制成作品。这样可以把浪费降到最低，并完全控制手绘在衣服上的效果了，相当于视频中展示的流程小规模实现。能数字化的原型制作步骤都已数字化。剩下的有些步骤还需要手工操作是由于：

1. 一个独立设计师工作室是买不起自动化设置的。工业打印机、烘干机、切割机都要$……$，没有像缝纫机那样的轻巧解决方案。

2. 可以外包给工厂，但开发速度太慢（见下文4）。要求设计师已经产生了大量的销售。如今我可以用3D模拟和手工制作的原型在网上获得订单。

3. 由于我设计->销售->制作，我希望在订单来了之后，每件都能适合每个客户。这需要在制版软件中进行手动调整（我使用Browzwear VStitcher）。

4. 即使我自己解决上述步骤，我给工厂提供了生产工艺单的时候，还会有很多问题。厂家使用不同的软件。每个工厂要求也不一样，设计师有不知道的具体细节。双方都浪费很多时间来调整工艺单以适应设施，上述工作重来。

5. 想保持零库存和大量量体裁衣，一件一件的做，以适应独特的客户的体形，每个订单进来时再马上做？2和4使得这样的想法无法实现。但这里描述的工作流程是最可持续和最具包容性的。

6. 视频中面料上的留白–演示很好，在生产中，这都是我们试图减少的废旧面料。我在Illustrator中对图案进行嵌套，以优化版与版之间的利用率。下脚料被用来做小东西（领子、头带、帽子、围巾、袋子、桌垫等）。但是厂家是不会为你做这些的。

7. 对制造商来说，获得批量订单的效率更高。使用排版软件来减少浪费和换工。但是，有人试过为不同的客户体型不同的样式放一块儿排版吗？

8. 根本没办法找到厂家只做切割和缝合，因为以上步骤的不顺畅。

这些是建立一个完全可持续的品牌所需要做的。我也希望看到供应商提供更多的环保面料选择，用于数字印刷和其他材料的可追溯性。总之，从设计到制造的每一步都应该有按需的大规模定制，并有道德的采购、升级和回收渠道。

有许多路径可以通向相同的结果。虽然圣杯可能无法完全达到，但在我们所处的经济体系中，对于可以实现的目标，人们可以通过建立一个企业来解决一些问题而定义世界如何演变。对于无法实现的、但对社会有益的理想，人们需要依靠写书来塑造未来的思维方式。

在这一过程中，约束数据和条件需要被优化，如年收入和商业杠杆。是方法上和思维方式的差异导致了我们所看到的竞争，有时甚至是敌意。如果目标是由道德驱动的，那同理心能让我们理解其实大家都想要同样的好结果和合作。