Full firepower, AI large-scale model strikes, is the turning point of this vertical field coming?

If you have experienced new home decoration, then home design must be the most troublesome part. Choosing a "wall clock" may be a "super project". From style to color and size, patients with difficult selection will directly collapse.

With the popularity of large-scale model technologies such as chatGPT and Midjourney, everyone found that this matter can be very simple.

Select a specific area, enter prompt words such as style and shape, and AI drawing will give N pictures with one click, and select a clock that is consistent with the style of the whole house.

Imagine if the entire home decoration link, from design to final purchase, is one-click drawing, what you see, what you get, this will undoubtedly be subversive for the entire industry chain.

In 2022, chatGPT will bring AI large-scale model technology into the limelight, and almost all first-line Internet companies will pour into the research and development of general-purpose large-scale models. Over the past year, the market has gradually discovered that the market potential of small and medium models in vertical industries is also very large.

"If your product calls an interface of chatGPT, then your product has no barriers, and everyone can do it. If you have domain data, proprietary data, no one else has it, and creating a small model will subvert a certain industry." Senior practitioner PlayerKang analyzed.

Invasion of large models, the collective opening of the home decoration design industry?

Will AIGC revolutionize the home improvement design industry? The answer to this question is found in Xiaohongshu. In Xiaohongshu, enter home decoration design + AIGC, and you can find a large number of beautiful home decoration pictures designed by AIGC. What's more interesting is that a large number of designer bloggers provide low-cost or even free home improvement design solutions. You give him a real picture + style requirements, and he can give 10 renderings in 5 minutes.

Are the 3 million home improvement designers in China trembling under the pressure of AIGC?

Today, not only Midjourney, but also domestic AI painting applications such as Unbounded AI and Chenyu AI can bring magical effects in the field of home decoration design.

Figure: The left side is the real picture and the design sketch, and the right side is the home improvement rendering drawn by Unbounded AI

Originally, when a designer came up with a drawing, he first had to use CAD software to draw a three-dimensional space drawing, then import it into modeling software, and then render it with a renderer... These steps not only represent professionalism, but also huge costs, including time and money .

AIGC is basically a game changer. Just as AI painting breaks the boundary between painters and beginners, AI design also allows ordinary people to show their talents in the subdivided vertical field of design.

According to Jcolor of Chenyu AI, in actual scenarios, designers use AIGC to make conceptual design drawings. When communicating with consumers, both parties can efficiently complete the communication of decoration plans. AIGC can be called a "marketing tool".

In China, there are two "giants" in the field of home decoration design, one is Sanweijia and the other is Kujiale. Their sensitivity to AI technology is actually beyond the public's imagination. For example, Kujiale established the AIGC laboratory with a high profile, researching AI design generation in all space fields including home decoration, commercial space, real estate and construction. The same is true for 3D Home. According to Cao Jian, its 3D product director, they have introduced AI technology to empower home decoration design since 2018. They have accumulated rich experience and data, and have launched a number of AI-related applications.

We found that the designed business model is clear and the effect is immediate, which is a typical traffic-type entrance. This has led to the emergence of many related applications recently. This gives people a feeling that AIGC and large-scale model technology have rolled up the entire home furnishing industry.

The vertical field takes the lead in making efforts. Will the field of home decoration design produce its own "iconic" model?

chatGPT set off a wave of AI large-scale model technology, and both overseas and domestic Internet giants have entered the general-purpose large-scale model track. However, judging from the implementation situation in the past six months, the small and medium-sized models in the vertical field may instead be "launched first".

The general-purpose large-scale model application is of course amazing. A lot of design content in Xiaohongshu comes from Midjourney, but if it is really going to be put into production, it will stop.

Home improvement design software such as Sanweijia and Kujiale often generate a 3D space plan, divided into two pieces of content, one is a rendered, very realistic bitmap, which can be viewed by consumers. The other is the corresponding 3D vector diagram, which has rich material, size and other data, and is used to connect to the production end for production. It is these vector data, including custom cabinets and other final products, that can be assembled seamlessly in consumers' homes, realizing "what you see is what you get".

What AIGC currently lacks is obviously vector data. How to overcome the "gap" between AI design and production? Cao Jian told Babbitt that this is a direction they are currently working on. Are there any challenges here? Cao Jian believes that the difficulty lies in the understanding and precipitation of the industry, and the emphasis on investment in new technologies.

Today, if the large amount of data accumulated in the home decoration industry is fed to the large model, and then the industry model in the vertical field is trained, so that the graph made by this model corresponds to the vector data one-to-one like the previous scheme, it is possible to realize the transformation of ordinary consumers. Become a designer's magical effect: not only the aesthetics are guaranteed, but also the production can be carried out.

The interesting point is that chatGPT cannot do this, and the Internet giants may not have this ability. It is reported that Sunway Home has a library of 1.63 billion+ design schemes, covering the household type data of 28 million+ communities across the country. Such a large and real vector data set makes it possible for the industry's large-scale model to be applied. Not only the data, but also the understanding of the upstream and downstream of the industry is also a huge gap in front of other entrepreneurs, which is difficult to bridge. Moreover, 3D experts believe that compared with the popular general-purpose large-scale models in the world, this vertical 3D large-scale model takes into account the security and controllability of data, and has the advantage that general-purpose large-scale models cannot penetrate.

In fact, Sanweijia is very fast, constantly exploring the possibility of AIGC landing in the home decoration design industry through self-research and cooperation.

On May 22, its Ai Design was upgraded and launched. According to its introduction, using AI technologies such as user type recognition, furniture recognition, and smart layout can solve the most difficult problem of opening blind boxes in AI painting, and realize AI creation with traces to follow , Creativity is controllable.

On May 15th, Sunway Home and Wujie AI reached a strategic cooperation, with a view to jointly applying AIGC technology to the home decoration design industry, and jointly exploring and creating a super large model belonging to the home decoration design industry through the algorithm technology of Wujie AI.

The question now may be that it is not when AIGC will empower home improvement design, but how far we are from the end of design software.

Picture: Various renderings produced by 3D HomeAi Design

AI is coming, do we still need designers?

"I think AIGC is unlikely to replace designers, but will accelerate their growth."

In Cao Jian’s view, the older the designer, the more valuable he is, and the lack of experience and experience will cause loopholes in the design plan, which will make the owner step into the pit. AIGC allows designers to spend less time on drawing, and instead he can learn more knowledge, such as strengthening the understanding of materials, innovation of construction technology, etc. Designers need to learn a lot of knowledge, and drawing is only a visual expression skills.

"AIGC has definitely had an impact on purely mechanized labor. Before the tools were developed, people became tools to perform mechanized labor."

From this perspective, AIGC does not replace designers, but instead promotes a large number of design practitioners to evolve from operators to designers.

However, the impact of AIGC is obvious. We found that in the home decoration design industry, the impact of AIGC will come faster in the following two aspects.

First, it affects the design efficiency of the interior industry and reduces the cost of the owner's design scheme.

Second, it will hit the work of graphic designers who mechanized labor.

In the future, the improvement of AIGC's efficiency will inevitably lead to a weakening of labor demand, which may be a challenge that has to be faced. But for individual design practitioners, in the face of the torrent of the times, the sooner they adapt to the trend and take the lead in arming themselves with tools, the only way to protect themselves from being overwhelmed by the waves of the times.