How an Israeli engineer went from Rafael to iPhone Portrait Mode and a $300 million OpenAI exit
Ziv Attar spent years developing camera technology before Apple acquired his startup in 2015. His second company, Glass Imaging, pursued a different idea: using AI to make smaller cameras capable of producing higher-quality images.
Few consumers are familiar with Ziv Attar, the Israeli entrepreneur revealed on Monday to have sold Glass Imaging to OpenAI for $300 million. Yet if you own an iPhone, there is a good chance you have used technology he helped pioneer, including the smartphone camera’s Portrait Mode. With the sale to the AI giant, Attar is cementing his position as a leading entrepreneur in photography and imaging.
Attar was born and raised in Afula and studied optical engineering at the Technion. After completing his bachelor’s degree in 2002, he began working as a senior optics specialist at Rafael. In 2007, he founded OpEra Optics, a startup developing optical and miniature camera technologies, where he served as Chief Optical Designer. In 2011, leveraging OpEra’s intellectual property, he co-founded LinX with Andrey Tovchigrechko.
LinX developed a compact mobile camera capable of producing high-quality images while being significantly smaller than conventional mobile cameras. Its technology relied on multiple image sensors and lenses rather than a single sensor. The design increased light sensitivity, as each sensor captured a different part of the image, reduced motion blur and enabled 3D photography.
Under Attar, who served as CEO, LinX also developed a dual-camera-based version of “Portrait Mode,” which keeps the foreground subject in focus while blurring the background. The feature would eventually become standard across the smartphone industry. LinX also developed an early dual-camera system for mobile devices that combined monochrome and color images to improve image quality in low-light conditions.
In early 2015, Apple acquired LinX for $25 million. Attar joined Apple’s camera algorithm team and helped turn LinX’s technology into a mass-market product: Portrait Mode, which debuted on the iPhone 7 Plus in 2016. Today, it is a standard feature on iPhones, but at the time it represented a significant advance in smartphone photography, helping narrow the gap between phone cameras and professional cameras.
“Portrait Mode is an example of computational photography,” Attar said about a year and a half ago in an interview with the CXOTalk podcast. “It’s not something that could have existed before we had sufficient processing power.”
Portrait Mode uses multiple cameras to generate a depth map, allowing the phone to determine the distance of objects from the camera. The system then uses that information to simulate the background blur produced optically by large camera lenses. “It isn't perfectly precise, but it is accurate enough to mimic the blur found in large cameras,” Attar said. “Portrait Mode was an attempt to bridge that gap.”
At Apple, Attar met Dr. Tom Bishop, and in 2019 the two left the company to found Glass Imaging.
“We want to revolutionize digital photography, particularly through AI,” Attar said. “It’s not about taking existing photos and enhancing them, that’s something many people are already doing. Our goal is to create hardware changes that are only possible thanks to the AI we are developing.”
The idea was to use AI not simply to improve photographs after they are taken, but to change what is possible at the hardware level. According to Attar, the technology could allow manufacturers to build smaller and cheaper cameras that deliver higher image quality, as well as thinner cameras that take up less space inside increasingly crowded smartphone designs. It could also enable new types of optics and camera-module architectures.
Attar argued that while AI had been integrated into smartphone cameras for years, its applications were largely limited to lower-resolution tasks such as facial and object recognition. That began to change as smartphones gained enough processing power to run AI on full-resolution images.
“In the last year or two, we’ve reached a point where AI, or the AI engines within phones, can handle full-resolution AI processing,” he said. “This opens up possibilities for highly innovative, impactful work, like what we’re doing at Glass Imaging.”
Glass’s approach involved comprehensively characterizing a camera’s optics and sensor, accounting for factors including individual pixels, viewing angles, types of light and light intensity. That data, which can amount to terabytes, is fed into software that trains a neural network capable of running on a smartphone. The system can then take imperfect images and generate higher-quality ones while correcting optical imperfections, sensor noise and other defects.
The approach could make it possible to achieve high image quality with less sophisticated camera hardware. That may help explain why OpenAI was interested in Glass Imaging.
OpenAI, working with former Apple design chief Jony Ive, is developing new AI-powered hardware, although the specifications and even the precise form of the product remain unknown. A camera could be an important component of such a device, particularly if it needs to understand and respond to its physical surroundings.
If Glass’s technology can help OpenAI achieve high-quality imaging with smaller or less sophisticated hardware, it could address one of the central challenges of putting powerful AI into compact consumer devices. That would give the acquisition significance beyond photography: it could provide OpenAI with a piece of technology for translating the physical world into data that its AI systems can understand.
