Cameras, a crucial factor to assess how good a smartphone is, have improved dramatically over the past decade. Facial recognition, filter effects, QR code scanning for payment: Tons of features rely on camera capabilities to support proper operation. Driven by the success of Pokemon Go, the rising VR/AR technology is now shaping future innovations for smartphone cameras.

Xu Yidan

Xu Yidan, CEO & Founder of Topplus (Image credit: Topplus)

For Xu Yidan, founder and CEO of Topplus, upgrading the smartphone camera is a great way to tap into the burgeoning VR/AR industry. Smartphones, due to their ubiquity, are proving to be the best use case for most applications.

“Smartphones will continue to record the largest smart hardware shipment for the next five years. Compared with helmets and goggles, it has user penetration. More importantly, the ubiquity of mobile payment services on handsets has made it a more accessible means to monetize VR/AR content,” he says.

The Chengdu-based computer vision startup develops camera SDKs and models that enable superior digital cameras functionality for supporting depth mapping, VSLAM (visual simultaneous localization and mapping, a key visual sensing technology), and multi-angle sensing.

The company’s MagicBar, an intelligent camera module embedded with depth function and VSLAM algorithms, enables smart devices to sense the environment comprehensively, and to build up a 3D map in real time.

Topplus’ developer kit allows developers to turn any smartphone camera into a Kinect-like motion sensor for AR games. By leveraging VSLAM and multi-angle sensing technology, a demoed fishing game showcased by the company allows up to four players to enter the same gameplay field simultaneously. Each gamer sees the same setting from different angles and all compete for the same prize-winning fish.

Elbowing into a highly compact market

Providing camera solutions isn’t something easy because it involves partnerships with all players along the smartphone manufacturing chain—camera model makers, chip makers and OEM factories.

“Smartphone manufacturing is a relatively conservative industry. Given the low margins brought by fierce competition, smartphone makers require camera solutions to be ultra stable in order avoid high repair rates. The market is even harder to penetrate because only those with at least 5 million shipment capabilities got the chance to enter the playground. No one wants to take the risk and to do ‘experiment,'” Xu Yidan pointed out.

Topplus’ entry point to the semi-closed market smartphone camera market lies on its patented technology for blurred portrait bokeh technology. Bokeh, a blurring background effect, depends on wide apertures to create depth maps with close-up subjects rendered in focus. After Apple introduced this feature to their phones, this effect has become more popular and could end up as a standard feature for every smartphone in the future.

“While Apple and Huawei hold their patents in the sector, they seldom sell it to smaller smartphone makers. We are licensing this technology to middle and lower-tier smartphone brands. Gaining legitimate patent license is a huge issue when Chinese smartphone makers try to enter overseas market,” Xu Yidan said.

What’s more important than the licensing fees collected from smartphone makers, the bokeh technology enables Topplus to reach deeper cooperation with companies on every link of the whole industrial chain: smartphone makers to open their plant floor data, chip makers for their workflow and model manufacturers for their various specs.

“This data is the prerequisite for being part of the industry, for providing updated camera SDK and model solutions. It’s also where we start,” Xu said.

Choosing the right direction is even harder for AI startups

Few startup ideas are perfect at the very beginning even for internet giants like ofo. That’s why it’s important to fine tune your direction along the startup journey. However, choosing a direction isn’t that easy. Every direction can be the right or the wrong choice simultaneously and nothing can be sure until it’s well underway.

This dilemma is particularly felt by AI startups since the whole industry is still in the early stage of finding potential application areas. “There’s too much of a bubble in the AI industry. Most of the AI startups are living on VC funding rather than generating sustainable revenue themselves through digging deep in a certain vertical,” Xu said.

“There’s too much of a bubble in the AI industry. Most of the AI startups are living on VC funding rather than generating sustainable revenue themselves through digging deep in a certain vertical,” Xu said.

With their computer vision software, one of the hottest branches of AI, Topplus has had its share of struggles. Before shifting to camera solution at the end of last year, the startup has previously focused on virtual glass try-on solution for retailers and tracking and image stabilization system for drones.

As someone who has experienced the painful process, Xu Yidan gives his definition of what’s “a good direction:” “If a direction has the potential to gain 10 billion RMB revenue in the long term, and you can survive on it in the short term,” he said.