How does AIGC change the 4 trillion home furnishing industry chain from rough housing to refined decoration in 1 minute?

Original source: Yibang Power

Author: Hu Weixin

Image source: Generated by Unbounded AI‌

Have you ever thought about designing your own home?

Just upload the photos of the rough house to the AI home design software, add descriptors such as "retro, warm, and suitable for a daze", and several realistic decoration renderings will appear in front of you. You can also adjust the color matching, lighting type, location type, and add furniture. "This process seems to be playing a game of dressing up." A user who has experienced AI home design said.

Image source: Kujiale AI

If the improvement of computing power and the upgrade of rendering engine in the past few years have made cloud-based home design software accepted by more and more individual users and designers, then the help of AI, especially the generative large model, has a great impact on design Subversion brings AI home design to a new stage.

At present, the application direction of AI in home design software is mainly reflected in three aspects: improving drawing efficiency, providing design inspiration, and intelligent error correction. With the deepening of the application of AI in home design software, not only is it expected to greatly reduce the threshold for designers, break through the growth restrictions of dealers and brand owners, but also present any demands of consumers in the design plan, and integrate with the back-end production Get through and truly realize personalized custom design.

According to Frost & Sullivan's calculations and data from the China Business Industry Research Institute, in 2024, the sales of China's furniture decoration and furniture industry are expected to exceed 4.83 trillion yuan, with a compound growth rate of over 20%, and the scale of the customized furniture market is expected to exceed 500 billion yuan , The compound growth rate exceeds 16%. Among them, the initial decoration of newly purchased rough houses, the redecoration of newly purchased hardcover houses, the renovation of newly purchased second-hand houses, and the renovation of existing houses are the main needs.

"Why are there 100 billion or 200 billion companies in the home appliance industry, but not in the home decoration industry?" Cai Zhisen, CEO of Sanweijia, issued a "question of the soul", "Home decoration is indeed very complicated, whether it is a shopping guide, whole house design, or physical modeling. , or the commercialization of materials, quotations, ordering, and order splitting. If you can use one drawing to connect all non-standard parts and standard parts, and complete marketing, design, ordering, production, and delivery in one stop, We can make furniture and home decoration a highly automated and intelligent industrial chain collaboration like the home appliance and automobile industries."

AI blessing, lower design threshold

Interior designer Yang Mingyu is a rookie in the industry. He runs a designer studio called "Wu Wo Art House". Before that, he was active in Hunan Satellite TV's designer variety show, helping Su Xing and Wang Lixin to do interior renovations. Yang Mingyu believes that AIGC can help designers save at least half of the time in negotiating orders with customers, making CAD drawings, and modifying renderings.

In the past, the designer needed to draw the structural drawing of the rough house into CAD through on-the-spot measurement, and this process took 1-2 weeks. Now, when a customer takes a few photos of the interior, or a designer simply sketches a few lines, the AIGC can directly generate a real-size CAD drawing, and a set of interior layout templates can be generated in one minute.

On the basis of the CAD drawing, Yang Mingyu also needs to make repeated adjustments according to the style, color, layout, lifestyle and other elements desired by the customer to determine the final decoration rendering.

Picture from: 3D Home

This process requires the cooperation of CAD+3DMax to render the renderings for customers' reference and modification. 3DMax is the standard equipment for rendering renderings. It takes one hour to render a picture. For a small detail change, such as changing an apple on the table to a pear, you need to wait another hour. Yang Mingyu said, "This process is irritating."

The AIGC tool does not have this problem. Change the details, change the background, and change where to point. AI can also actively provide design inspiration. Coupled with a high-performance rendering engine, it can render 4 new design drawings in one minute.

Image source: Wu Wo Art House

With the AIGC tool, Yang Mingyu's project time was directly halved. "Now the CAD can be completed within a week; the communication and modification of the renderings, the small three-bedroom and large three-bedroom of 80 to 100 square meters, can be completed within one month. Then enter the construction stage."

For furniture manufacturers and custom manufacturers, AIGC is not only suitable for product design, but also facilitates store sales and online marketing.

The decision-making cycle of home improvement customers is as long as 30-90 days. Offline stores are still the main channel. In order to convert customers who enter the store into orders, brand owners have racked their brains.

According to Sunway’s 2022 survey of nearly 100 soft furnishing stores, 26% of the shopping guides choose to show customers the details page of Taobao or the product’s official website when faced with customers who need renderings, and 63% of the shopping guides guide customers on-site. To experience the real thing, less than 6% will make renderings for customers. To this end, Sanweijia launched a 3D show. The shopping guide only needs to import the house type, and the other processes are completed by AI independently, and a VR panorama can be generated in 10 minutes.

"CAD software requires learning costs. In the past, although our cloud design was much more intelligent than AutoCAD and 3DMax, it was not enough to be used out of the box. I think this round of AI is to allow our users to use it with zero barriers." Cai Zhisen said.

Image source: Sanweijia

Every designer is based on the 3D virtual live broadcast room, 3D model room, 3D exhibition hall and other virtual scenes built by AIGC, highlighting the diversified immersive scenes of the color, material, size, style and other commodity characteristics of furniture. For the content production of short videos and pictures, the designer of every flat every house launched a 3D studio shoot for popular products. From the selection of studios, product changes, selection of shots/paths to submission of renderings, all are completed by the platform. A "big movie" It only takes a few minutes to generate.

Kujiale also launched the AI studio shooting function of Kujiale Studio Shot 2.0, which forms an artificial intelligence commercial shooting platform based on Kujiale's original 3D rendering technology and massive materials. No need to build a scene, cut out, compose, adjust light and shadow, just upload a product picture, AI will automatically cut out the picture, and automatically generate a multi-scene product picture according to the style and element selection.

Image source: Kujiale Studio Shoot 2.0

Chen Hang, CEO of Qunhe Technology (Kujiale), believes that pan-furniture is a service-oriented industry that penetrates consumers through services, and the service process is often time-consuming, labor-intensive and labor-intensive. Humans plus AI will form stronger service capabilities and popularize the best services to everyone. This is a huge opportunity.

From Wen Sheng's "Picture" to Wen Sheng's "Room"

The next question is, can the pictures made by the design software really be produced? Can it be decorated exactly the same?

Yang Mingyu understands the client's concerns. In the field of design, especially when it comes to home design related to physical objects, the biggest problem is the wrong version of the picture. Interior designers need to achieve what you see is what you get, "It's like eating instant noodles, and there's a lot of beef on the package, but it's not necessarily obvious after you open it." Yang Mingyu added.

This is also a difficult problem for the general drawing model.

Chang Jia, the founder of Unbounded AI, found that interior design has reasonable requirements for lighting and space layout. Lighting and shadows must conform to optical principles, the location of light sources must be accurate, and indoor layout must conform to the industry's normal design specifications. At present, most of the AI pictures can meet the requirements of style and aesthetics, but they are lacking in the rationality and logic of the pictures.

The images generated by Midjourney and Stable Diffusion focus on being out of nothing, novel and interesting, but they cannot really enter the production line.

Qian Min, the manager of "First-line Design", has used AI drawing tools for half a year. He believes that Stable Diffusion and Midjourney are more suitable for creative design. The biggest problem with interior design is that the details, size, and layout of the scheme will change accordingly. Variety. If you want to restore it according to the true proportion of the actual size, the design time will increase a lot and the efficiency will be relatively reduced.

"Midjourney randomly generates a living room with a height of five or six meters. It looks beautiful, but the actual thing is not that high. Stable Diffusion can control the scale, but the beauty is not as good as Midjourney. In short, the higher the randomness of AI, the harder it is Control." Therefore, Qian Min combined multiple design tools such as Stable Diffusion and Midjourney.

Image source: Line Design

How to train an AIGC tool that can be used by the home furnishing industry? Cai Zhisen, CEO of 3D Home, believes that the core lies in the vector data set.

A vector, also known as a vector, refers to a quantity that has both magnitude and direction. Vector data generally expresses the spatial position of an object accurately by means of coordinates. "The design drawing that the user sees seems to be a plan, but all the tables, chairs, benches, wallpaper and floor tiles, sink and TV cabinet in this picture have relative positions and sizes, and they are all vector data." Yang Mingyu pointed out.

Image source: Sanweijia

For example, a physical sofa, viewed from top to bottom, from left to right, has fixed dimensions and lines. At the same time, the size and color patterns of the sofa can be scanned into a physical asset, a vector model, and the designer is calling These physical assets are designed. "You can take a photo of an interior and generate a brand new design drawing through AIGC. The shape and material are still controllable. This is the role of vector AI, with its own production data." Cai Zhisen introduced.

Behind the vector dataset are massive industry data assets.

For example, Sanweijia has more than 20 billion home decoration model parameters and tens of millions of real cabinet data sets. "Now our housing data is on the order of 1 billion. There are at least 1,000 parts in each house. The attribute label of each part is also 100. Every day, about a million houses are generated through cloud design." Cai Zhisen express.

Image source: Sanweijia

Based on these industry data, in July, 3D Home launched the "Whole AI Brain" (3D Home vector AI model). The vector AI large model can be used for the AI design of the whole house space, the AI design of a certain wardrobe and cabinet, and can also generate a new design drawing based on photos or line drawings. "Our AI model still has a lot of scalability, because AI is a recommendation algorithm, which can link regional parts and national parts to form a national link ecology. We can customize the whole house for customers." Cai Zhisen introduced .

At the Taobao Tmall Home Improvement and Home Ecology Conference on August 14, 2023, the designers of every flat and every house launched 3D digitalization to address the major pain points of the traditional home improvement industry such as high cost, long cycle, difficulty in changing products, and lack of inspiration. Product matrix, including 3D scene-changing live broadcast, 3D explosive studio shooting and global marketing platform.

The person in charge of the designer of every flat and every house once said that the designer will implant AIGC application capabilities into the open platform of Taoxi, and integrate the design of every flat and every house with the home improvement industry of Taoxi in AI smart design, 3D smart scenes, electric appliances, etc. The underlying digitalization of merchants and stores, global content marketing and other technical capabilities are connected to the home consumption scene.

Qunhe Technology, which has accumulated more than 270 million product material models, has also released Kujiale AI, which can generate decoration renderings based on real-life pictures, creative solutions from floor plans, and AI one-click replacement of product scenes. Not limited to decoration scenes, Kujiale's AI capabilities are also embedded in more sub-sectors such as customization, doors and windows, building materials, as well as more space scenes such as office and catering. "Kujiale AI" launched a customized style model for designers and enterprises, which will be launched soon.

In addition to a large number of commodity material models, Qunhe Technology has also covered 90% of new house types in the past five years. At present, InteriorNet is the most widely used data set for deep learning of indoor scene cognition, and the total number of rendered images on the platform exceeds 3.3 billion.

Image source: Kujiale AI

Chen Hang, CEO of Qunhe Technology, once pointed out that the application scenarios of Kujiale AI are divided into three categories, the infinite inspiration map generation for owners; the precise and controllable AI creative design for designers; the AI replacement selection for brand store shopping guides . "Consumers, designers, and enterprises in the home furnishing industry are undergoing productivity changes. In the future, the 'heterogeneous computing' formed by humans + AI will bring further breakthroughs in consumer experience." Chen Hang concluded.

** How does the manufacturing industry realize "one picture to the end"? **

Although AIGC is easy to use and doubles the efficiency, Yang Mingyu found that many interior designers don't care about it.

"After all, the list in everyone's hands is not finished yet, so don't worry about AI." Yang Mingyu believes that the daily work of home decoration designers is too saturated, which makes many designers not sensitive to changes. The Shell Research Institute found that in addition to drawing plans, it is also necessary to communicate frequently with customers and partners, and to deal with various emergencies at the same time. "Frequent overtime work", "long working hours", and "less rest time" are the main slots.

At the same time, the impact of AIGC on interior design is far less rapid than that on graphic design and fashion design.

Designers include graphic design, fashion design, product design, interior design, and architectural design. Designers in different industries are impacted differently by AIGC. Yang Mingyu felt that graphic design was the first to be impacted, followed by fashion design, while the influence of interior design and architectural design was relatively lagging behind. "After the fashion design has revolutionized its own life, it will be our turn to revolutionize our interior design." Yang Mingyu thinks.

Data source: Boston Consulting Group

But for design software manufacturers and brand owners, many people realize that this is an extraordinary development opportunity.

Cai Zhisen, CEO of 3D Home, still remembers the impact brought to him when the big model was first released, "We have been working on AI, but everyone is constantly investing in it while doubting life, suffering repeatedly, and the two worlds of ice and fire. ChatGPT is first and foremost for practitioners. Confidence, let us invest in AI more resolutely." Cai Zhisen believes that in the long run, AIGC will bring triple benefits to home decoration design.

**First of all, AIGC accelerates the iteration speed of design products, enabling the iteration of SaaS products to reach an update speed similar to Moore's Law. **

**Secondly, AIGC addresses the ease of use of design tools. **

**Third, the popularization of design tools will accelerate the industrialization of the whole process of the furniture and home improvement industry. **

Lower usage threshold means higher industry penetration and more revenue. Cai Zhisen said: "For design software in vertically segmented industries, the market share is a few thousandths. I think this time AI should accelerate us to quickly push the market share to double digits, and at the same time, it is expected to break through faster. 20%, 30% market share. At the same time, breaking through the double-digit market share means that SaaS manufacturers can break the curse of 1 billion revenue.”

In the past, the product drawings of the furniture and home decoration industry chain were in different formats. "In the past, designers used 3DMax for renderings, AutoCAD for construction drawings, AutoCAD for factories, and Excel for BOM tables for quotations. Brand owners bought a bunch of ERP and didn't need it." , It is very painful to cooperate. If based on the same software, you can place orders and disassemble orders, make renderings and control machine tools, then the collaboration speed of the industrial chain will reach a new high.” Cai Zhisen pointed out.

Furniture and home improvement brands also saw an opportunity to improve the management radius and break through the growth bottleneck.

The industrial chain of the home furnishing industry is lengthy, the products are diversified, the demand is highly personalized, and the capabilities of service providers are not standard, resulting in a low market concentration and forming the industry characteristics of "big industry and small company".

Under the integrated sales trend of "whole store customization", the back-end production determines the delivery efficiency, installation success rate, and whether "what you see is what you get" can be realized. In addition to the AI application on the design side, the integration of front-end and back-end systems and back-end production Process intelligence is also the focus of manufacturers.

Image source: Sinolink Securities Research Institute

For example, the CAXA software of Oupai Home Furnishing integrates design renderings, quotations, wireframes, order placement, and automatic order dismantling. The back-end link saves nearly 6 hours compared with the previous one; the Sofia factory adopts CNC digital control system, which can process faster Production process data, and make production forecasts, improve the board utilization rate, effectively reduce the waste of wood resources; at the same time, the robot board sorting rate is nearly 100%, the accuracy rate is at the leading level in the industry, and the loss in the manufacturing sorting process is controlled .

Through the integration of all categories and the whole process, brand owners can quickly match the front-end design scheme with the back-end automated production data, avoiding the inefficiency and error problems of manual dismantling, allowing orders to enter the production process faster and improving overall production efficiency.

Some industry analysts believe that as the empowerment of AI in home products, front-end scheme design and back-end production continues to deepen, leading home furnishing companies are expected to rely on capital and technological advantages to further expand their own advantages in the AI era and break the original Due to the growth bottleneck caused by manpower constraints, the share continued to increase.