Whiplash Report Recently, the popularity of virtual anchors has continued to heat up.
Liu Yexi, a beauty expert known as a phenomenon-level virtual person, rose millions of fans in the first video, and debuted at the peak; with the "voice actor" debut advantage, the vibrato anchor Xu Anyi, who quickly carded virtual live broadcast, rose nearly 100,000 fans in more than three weeks, triggering a boom in onlookers. Earlier, Luo Tianyi, Ayayi, Tsinghua student Hua Zhibing... More and more virtual digital people are being sought after and are quickly becoming popular.
Research data show that the current virtual digital person market size has exceeded 200 billion yuan, is expected to reach 270 billion yuan in 2030, virtual person track is expected to emerge more than 5 unicorn enterprises. Among them, the identity virtual people represented by virtual anchors, celebrity virtual doppelgangers, brand spokespersons, etc. will occupy a dominant position in the future, with a volume of 175 billion yuan.
Virtual people, with the help of the east wind of the meta-universe infrastructure, suddenly stood on the cusp of the storm. Especially in the MCN and anchor markets with serious internal involvement, the emergence of virtual people has solved problems such as IP stability, reluctance to appear on camera, and differentiated competition, and is even considered a tool for anchors to counterattack. When it comes to virtual people, in addition to the image design modeling and rendering of the content side, the motion capture technology that determines the flexibility, stability and threshold of use of virtual people is a must for soldiers.
Laser + inertia: easy to build, easy to carry, continuous live broadcast does not "drop the line"
On April 17, The live broadcast room of Xu An, the virtual anchor of Douyin, was full of people. "To ten thousand people to give everyone a handstand, you can also dance at will, I will also handstand wash your hair, there are many unique jobs", online data quickly soared to 10,000 people, Xu An did not say a word or two, on the spot came a handstand, the whole action flowed, without any delay.
But it is precisely such a "conventional" action of offline live broadcasting, if the virtual digital person is allowed to "do it", the requirements for motion capture technology and equipment are quite harsh.
With the "laser positioning + inertia" motion capture technology, Xu Anyi "has no fear". Because before the use of STEPVR's motion capture program, the virtual anchor is easy to "drop the chain" in front of the window, or the image collapses suddenly, or the large-scale action "wears the gang", so it has to be cautious. With the blessing of "laser + inertia" motion capture technology, the anchor can perform arbitrarily in the live broadcast room, dance, handstand, flip head and other difficult actions, can be smoothly completed, but also support multi-person real-time action capture, ornamental at once pulled up a lot higher.
Compared with other types of motion capture technology, "laser + inertia" also has the advantage of super stability, continuous live broadcast for 10 hours, will not pull the crotch, do not need to correct the reduction and other unnecessary operations. In addition, it is worth mentioning that this motion capture technology is extremely cost-effective, the threshold for use is extremely low, and the space requirements are also very low, if the anchor wants to change a "field", carry the box out, in any small space can be flexibly built, live broadcast, almost fool-style operation.
Inertia: The cost is low, but the shortcomings are obvious and take 15 minutes to correct
At present, inertial motion capture technology applied to virtual live broadcasting is dominant. In principle, it is to apply the inertia sensor to the data acquisition end, process the data through the inertia principle, so as to complete the attitude angle measurement of the moving target, which can be simply understood as the gyroscope in the mobile phone.
The advantage is that the cost is relatively low, the short board is also very prominent, the error is relatively large, and the repeatability is relatively low. Because motion capture data is extrapolated, the accuracy of absolute position data is very low. For example, in the real world talent show, after the anchor returns to the origin, after the virtual world follows the same movement, the inertial virtual person is very likely to not return to the original origin, or even run to the beginning. If the data error continues to accumulate, it is necessary to reset the corrected device every 15 minutes.
For virtual anchors, this is embarrassing and often unbearable.
Therefore, the virtual anchor who uses a single inertial motion capture technology, whether it is dancing or exercising, the upper limit time is about 15 minutes, and then the anchor has to return to the seat to "rest", in fact, in order to reset the correction. In addition, when using inertial motion capture, if there are more peripheral mobile phone devices, the electromagnetic signal is complex, which is likely to lead to a sudden collapse and loss of control of the virtual person. At present, there are Xsence representative enterprises abroad, and there are many domestic motion capture enterprises that adopt inertia.
Optical camera: film and television level effect, complex construction, poor mobility
The last category is optical camera motion capture technology, which is not unfamiliar and has a wide range of applications in Hollywood cartoons such as Avatar. The technical principle is to cover the indoor space through multiple infrared emitting cameras, place reflective points on the tracked object, and determine their position information in space by capturing the images reflected back by these reflective points.
The advantage is the film and television level effect, because of the years of technical precipitation, the visual effect of the film is delicate. Of course, this is naturally the goal pursued by all virtual anchors, and the quality experience is more friendly to fans, but unfortunately, the price is extremely high, which is prohibitive. And the operation is quite complicated, requiring a long period of training, can not achieve rapid construction, after the completion of the construction can not be easily "moved", poor mobility, the need for professionals to maintain regularly.
But this is a bit contrary to the future trend of virtual anchors, who need more freedom.
For example, voice actor anchor Xu Anyi used virtual people to take a dark horse posture; after Liu Yexi exploded, he launched a non-stop offensive offline to receive brand endorsements; McDonald's launched the first virtual spokesperson, and the ultra-realistic digital person Ayayi also became a big brand cooperation household, and the future virtual people went to the scene of carrying goods. This means that the future scenarios for virtual people are more diverse, not only existing in the live broadcast room, which requires the dynamic capture technology behind it to have flexible, mobile, fast construction, and maintenance-free characteristics. Obviously, the drawbacks of the optical camera school are obvious, and some are difficult to adapt.
After a comparison, which is stronger or weaker than the motion capture technology? The results may already be self-evident: inertia has a certain cost advantage in the short term, but in the long run, the future may be the world of "laser + inertia" technology.