By LIN Teng
Chinese smartphone maker Vivo has revealed its research and development achievements in AIGC algorithms for the first time. The new V3 imaging chip supports multi-concurrent AI perception.
Vivo has established a 3D-motion-capture laboratory to find ways to apply next-generation AI to its products. Located in Vivo's Chongqing intelligent manufacturing center, the laboratory will play a crucial part in Vivo's intelligent manufacturing strategy.
In the lab, Vivo can construct "pore-level" 3D human models, with a focus on face, body, and gesture algorithms.
Vivo collects algorithm-learning data in the lab to solve imaging problems that are difficult to address with hardware alone. This includes, for example, background blur effects for portraits at the hair-strand level, facial depth, and texture for features like hairstyle or age editing. There are currently over 100 staff working on these subjects, most with doctoral degrees.
As AI becomes more widespread, mobile phone companies are accelerating their technological investments. Apart from Vivo, Honor proposed on-device in July, and Huawei will incorporate a large language model (LLM) in HarmonyOS 4.
Imaging has become the testing ground for LLM. Overseas, Midjourney and OpenAI have made breakthroughs in applications such as drawing and image generation. In China, a recent AI photo app called "Miaoya Camera" attracted a large number of users due to its simple operation, low price (9.9 yuan), and high-quality effects.
According to LI Zhuo, director of Vivo's imaging products, optical technology faces challenges since its analog and breakthroughs may require advancements in materials science and basic physics. However, once overcome, it could lead to a whole new world. Computing power and algorithms are showing speedy progress.
