A food creator wants to slowly push the camera from the panoramic view of the dining table to a bowl of noodles. This action is not difficult, but the difficulty is that he does not have a second person to help him push it, and there is no space to set up a slide rail on the table. Finally, he cut a fixed camera position. The picture was clear, but it lacked the sense of breathing.

This table is the real scene for many content creations today. Product reviews, food, beauty, live streaming, product delivery, a large number of shootings all take place within a few square meters.

In fact, the equipment is not bad: the mobile phone can shoot 4K, the pocket gimbal can stabilize the picture, and the picture quality of the camera has already exceeded the needs of most people. But once the lens needs to "move" and needs to be pushed, pulled, circled, raised and lowered, the set of things in ordinary people's hands cannot be connected.

Cameras are good enough, but moving a lens is still a craft that requires years of training. What Yueqian Chuangxin does is to turn it into an action that a machine can perform.

The founder, Zeng Cheng, worked on a film crew and studied robot trajectory optimization during his master's degree. The company was established in 2023. It first made large-scale photography robots for film crews. Later, it lost weight along the way and switched its technical route to a five-kilogram desktop product.

What supports this switch is the precise understanding of constant loads in dedicated scenarios: the weight of the camera as an end load is constant. After leveling, the gravity is balanced. The motor's job is not to lift the camera, but to move the camera. From this, zero-gravity direct drive technology is derived, which is also the basis for lightweight products. The combination of full freedom and AI director decision-making forms the product path of Yueqian Innovation - the first generation serves professional creators, and the second generation moves towards the consumer level, and may enter the price range of DJI and Shadowstone.

The imaging industry is reaching a point where new answers are needed. The ceiling of parameter competition has appeared, and the portable experience has reached its limit. The next stage of competition is no longer about "shooting clearly", but about helping users truly complete the content they want to shoot.

Why should a start-up company with 30 people take away a piece of cake from a giant? Both DJI and ShadowStone are talking about "director machines". Why does Yueqian Innovation start with a table?

In the following conversation, Zeng Cheng gave his answer from the aspects of technology selection, product definition, industry judgment and competitive landscape.

(Zeng Cheng, founder of Yueqian Innovation)

Leifeng.com: How did Yueqian Innovation first get involved in the field of imaging? What pain points did you see at that time?

Zeng Cheng: My master's degree topic is robot trajectory optimization. Then I worked in self-media and participated in behind-the-scenes work of the crew. While understanding the robot technology, I can also understand the knowhow and pain points behind the shots of different shooting scenes.

At that time, I saw on the set that the rent for a photography robotic arm costs tens of thousands of yuan a day, and two engineers are required to program it on site. The equipment needs to be reserved in advance and pre-programmed before shooting. It may take several hours for the director to go to the scene to change a requirement. The efficiency is low and the price is expensive.

At that time, we made a product, mainly by optimizing the operating software to optimize the original 6-hour workflow to 1 hour. But while doing it, I discovered that the ceiling of this road is not in the technology, but in the scene. The price and size of large equipment determine that it can only serve a small number of people. But the capacity of this market and the real increase in content creation are not on the same order of magnitude.

Leifeng.com: Where is the real increase in content creation?

Zeng Cheng: For individual creators and small teams. Because the way images are spread has changed. Ten years ago, it was traditional integrated media and TV advertising, but now many are disseminated by self-media.

There are more than 200 million active content creators in the world, and more than half of the shooting takes place indoors and on the desktop, such as unboxing reviews, food, makeup, live broadcasts, goods delivery, etc. They have creative needs. Their lenses need to move and the pictures have a sense of breathing. However, current equipment cannot meet the needs well - the equipment in their hands is mainly slide rails and stabilizers, which have a single degree of freedom and a time-consuming process. A small number of top creators can use millions of large devices, but most creators’ workspaces simply cannot accommodate large devices.

So we redid the entire technical route, from a single degree of freedom to a full degree of freedom, and from large equipment of hundreds of kilograms to desktop level.

Leifeng.com: From hundreds of kilograms to desktop grade, what exactly did you go through?

Zeng Cheng: In fact, the earliest prototype was an industrial robot plus a base, weighing 500 kilograms. Later, I switched to collaborative robots and the weight dropped to 120 kilograms. Then use carbon fiber and replace it with a smaller motor, 70 kilograms, 40 kilograms, all the way down.

When I reached forty kilograms, I realized that following the weight reduction logic of industrial robotic arms, I could no longer walk. Its structural logic and load logic are designed for industrial scenarios such as grabbing and handling, not for shooting with cameras. If we don’t make breakthrough designs, we can only get there at most. Performance, size, price, none of the indicators can get us to where we really want to go. So starting last year, we radically redesigned the entire technical route.

Leifeng.com: Route switching and weight reduction mean giving up something. What did you give up?

Zeng Cheng: The no-load scenario was abandoned. When you take the camera off, the robotic arm becomes unusable. Just like the gimbal logic, you need a camera to hold it down first. We also give up a certain amount of precision. The accuracy of the second-generation robot arm is 5 filaments, which is the accuracy of half a hair. When moving to the same place twice, the error is no more than a hair. But shooting does not actually require such high precision. Instead, it requires continuous movement. This is a trade-off we make.

Leifeng.com: How can we ensure the continuity of filming?

Zeng Cheng: In the industry, six degrees of freedom are called full degrees of freedom. Six degrees of freedom can achieve any position and any attitude. However, six degrees of freedom will have singular points. When the robot arm moves near certain configurations, it will be unable to continue and must go around. Industrial scenes emphasize the "result" of movement. It doesn't matter if you go around or stop for a while, but shooting emphasizes the process of movement and cannot be interrupted during the shooting, so we need to solve the problem of singular points.

We have designed a special configuration, which can be combined with a self-developed control algorithm to cleverly avoid singular point problems during camera movement, allowing the trajectory to be completed continuously.

Leifeng.com: There is an answer to the question of how the machine moves. So back to the imaging products themselves, what is the iteration logic of this industry over the years? Where will the next stretch of competition take place?

Zeng Cheng: I divide imaging products into three stages. The first stage was the era of Canon and Nikon. Everyone competed for image quality and who had better parameters sold better. In the second stage, after the parameters reach a certain threshold, if users continue to increase the parameters, they will no longer be able to perceive them, and the industry will start to experiment. For example, pocket cameras, a category pioneered by DJI, can provide a good portable experience and produce good pictures. In the third stage, it is no longer meaningful to continue to upgrade the parameters of pocket cameras. So I think the third stage is to use AI and robots to ultimately improve the user experience of getting the film.

Leifeng.com: Why “filmmaking experience”? What are the limitations of current equipment?

Zeng Cheng: Take the video footage captured by Movie Hurricane with Pocket as an example. A novice would not be able to capture the same video with the same equipment. The flow, breathing, and overall rhythm of the picture are beyond the reach of ordinary people. We have studied the reasons in depth - professional creators like Movie Hurricane know the three stages of creating a video, and can execute it at a high level.

Ordinary users often do not have a complete creative process. There was no planning before shooting, so I shot whatever I saw. During the shooting process, I didn’t know how to choose the focal length or compose the shot. In the end, I shot a hundred or two hundred pieces of footage, but I couldn’t cut them out. This is actually the key point - now the emergence of AI plus robots can solve this problem.

The products currently on the market are either pure photography or pure generation. What we focus on is to put AI's capabilities at the highest value point to help users take photos. AI is responsible for the directing part - AI acts as the director, the robot acts as the photographer, the user participates in the process, and the final product is a good finished film.

Leifeng.com: How exactly does this “AI director” work?

Zeng Cheng: In fact, AI directors are not the key to the problem. The director's intention must be implemented to ensure the quality of creation. AI can complete the creation of the script, while the execution of the script is completed by robots, and the final editing is completed with one click using AI. This is roughly a workflow. Product capabilities should run through the entire early, middle and late stages of creation, rather than the intelligence of a single local function.

Leifeng.com (public account: Leifeng.com): What is the product planning of Yueqian Innovation?

Zeng Cheng: There are two generations of products, and the user groups of the two generations are different.

The first generation is for professional creators and mainly serves the mid-term implementation stage of professional creators. The real scarcity of this type of creator lies in motion control, not lens creativity. The second generation will be smaller and can be used as the main camera for novice users, and can also be used for shooting in movie style.

The first generation can be understood as putting Hollywood on the desktop of creators. The second generation puts the language of film in your pocket.

Leifeng.com: What is the approximate form of the second-generation product?

Zeng Cheng: The second-generation product will be more portable and easier to use, and will not require users to be so professional. There are also many technologies in the middle that are being verified, and they are currently classified as confidential internally.

Leifeng.com: Why go into the professional track first instead of going directly to the consumer level?

Zeng Cheng: There are two reasons. The first is brand. Whatever tone you set for your first product will determine where users will place you. Let the most professional and picky people use it first. While capabilities are accumulated, the brand also has an anchor. The experience learned from professional users can also be given to consumer users, improving the experience while also increasing user trust.

The second one is data. Professional creators know what a good lens is, when to push it in, when to circle it, and what focal length to match with what scene. These judgments are real creative decisions and cannot be written in a parameter sheet. In the process of serving them, the first-generation products will accumulate these decision-making data and use them to train the intelligent capabilities of the second-generation products.

Therefore, the relationship between the two generations of products is very clear: the first generation serves professional users and is responsible for maintaining the brand and collecting capabilities; the second generation targets ordinary users and translates professional decisions into shooting plans that can be directly called.

Leifeng.com: What types of customers are you actively seeking now?

Zeng Cheng: There are several categories. One category is the BBC, Hollywood, CCTV, etc.; the other category is self-media bloggers, MCNs, and the company’s internal e-commerce team.

Previously, the photography director of CCTV9 approached us and wanted to shoot a cultural relics documentary. Since the cultural relics cannot be moved, they had to film on site. They drove two trucks into Tibet, and the equipment their current supplier gave them required four large flight boxes, which took up half of their truck; it could fit in one of our suitcases. He came to see us several times, but it was still in the early stages of technical verification at that time and he did not catch up with their project. He said we must tell them when we go out, and they will allocate one to each group. The BBC also sent an email to ask.

The other category is the company’s brand team. Any company with its own products and brands that wants to be an e-commerce company must have its own photography department, and the equipment can cost tens or even millions to buy. This type of user has the need to continuously create content.

Leifeng.com: What unique technical accumulation do you have in operation control?

Zeng Cheng: Yueqian Innovation’s operation control is specially developed for scenarios. The experience of using it only needs to make users feel comfortable. In the past, there was no way to directly use motion control in industrial scenarios, and it took several hours to edit a trajectory program. We're going to shorten this process to just a few minutes. In order to achieve such an experience, special design is required from the body to the operation control. Combined with user feedback after the product is released, we will iterate and optimize again and again, and we will naturally gain more technical accumulation.

Leifeng.com: The second generation products may directly compete with DJI and ShadowStone. What do you think of this competition?

Zeng Cheng: The market for imaging products is large enough and can accommodate more companies. Just like Apple's mobile phones are doing well, they haven't eaten up all the mobile phones on the market.

User needs are not a point or a line, but a space composed of countless points. No company can meet all demand points and must make trade-offs. The points that are discarded do not mean that they have no value, but the value needs to be concentrated on the points that are left behind. The opportunity for startups is to pay attention to different value points from large companies, make different trade-offs, and thoroughly understand the demand points that remain in the choice, so that they can also make excellent products.

Leifeng.com: How big do you think the imaging market is? What are the unfulfilled scenarios?

Zeng Cheng: It is predicted that the market size of consumer-grade smart imaging hardware will not be less than 100 billion.

The imaging market is so huge that it occupies the two most important senses of the human body, vision and hearing. Images and sounds can accept the most information and come in various forms. There are so many needs that can be tapped, and there are still many unmet needs. For example, in sports scenes such as ball games, the World Cup, and tennis matches, those viewing angles cannot be achieved by consumer-level users, but that viewing angle is the most beautiful viewing angle and can record the highlight moments.

Leifeng.com: What is your initial estimate of how many people have entered the video circuit now? Who is the real opponent and who is a passerby?

Zeng Cheng: The imaging track is large enough and will naturally attract many talents and entrepreneurial teams who want to find some new opportunities. It may first require that the entrepreneurial team be able to find the real user needs, deeply understand the scenario, define the product well, and be able to do a good job in product delivery and after-sales. Teams that can achieve these points should be qualified opponents. As for who will be a passer-by, users will vote with their feet.

Leifeng.com: From the demand side, what is missing in current imaging products?

Zeng Cheng: Professional users lack easy-to-use creative tools, while consumer users lack a complete creative flow. What Yueqian Innovation wants to do is to productize creative capabilities starting from professional scenarios and provide users with an intelligent product with complete creative capabilities.

Leifeng.com: In the further future, where will images go?

Zeng Cheng: The essence of human creation of image content is for expression. Whether it is photographed or generated, as long as it can help expression, it will remain. Only imaging products that truly serve content production and help creators express themselves better can survive.

Most of the progress in imaging hardware over the past ten years has occurred on the "recording" side: higher image quality, smaller size, and more stable anti-shake.

From "What do I want to shoot like" to "I really shoot like this", between creative intention and camera movement, there is a whole set of abilities that require long-term training to acquire.

An untrained person, even if he holds a good enough camera, still doesn’t know where the shot starts, where it goes, and where it stops.

What Yueqian Innovation wants to do is to dismantle and productize this set of capabilities, and break down professional judgment into steps that can be executed by machines and invoked by users.

If this path is taken, it will not only change the shooting equipment, but also the definition of "who is qualified to take good pictures". Whether this path is established, the market will give the answer after the product is launched next year.

But it at least points to an issue that has been ignored for a long time: when the equipment itself is no longer scarce, what is scarce is the ability to use the equipment to achieve results. Whoever can make this ability callable may find his or her place in the next period of competition.

(leifeng.com)

Related articles:

Why is the leader in overseas smart hardware betting on this AI-native marketing platform?

Multi-category development, AI internalized in the core: Anker Innovation demonstrates the strength of the global leader with an intelligent hardware matrix

The next step for AI is “action”: YoooTek debuts at the Smart Hardware Exhibition with its first smart hardware, AI ONE

The winner of the Bund Conference AI Science and Technology Innovation Competition is announced! Nearly 20,000 people competed for the first, second and third place in the three tracks of intelligent hardware, AI security and financial intelligence.

Original articles from Leifeng.com, unauthorized reproduction is prohibited. See the reprint instructions for details.

edit

Resend Email Modify Email