Q&A: A phone photography engineer on aiming for DSLR quality, future improvements
Photography is an obsession among phone-makers. And they have reason to obsess on it because photos are among the most tangible ways that most consumers see how much a new phone has improved over last year's models. There's always some magic in seeing an amazing photo taken by those tiny cameras at the back. It's an engineering feat.
We recently got the chance to have a quick email Q&A with one of the behind-the-scenes men tasked with figuring out how to make phones take increasingly better photos. Will Lin is a photography algorithm engineer at OPPO, a brand that, over the years, has certainly made headway in competing with the likes of Apple, Samsung, and Huawei.
Lin gives us a peek at where OPPO's phone photography is headed, including their new "Ultra Night Mode" and "Ultra Dark Mode" – lowlight technologies that they believe could go up against the best there is in the industry.
Q: You are a photography algorithm engineer. Can you explain to us what you do, and what you and your team’s objectives are in OPPO?
Will Lin: I am leading an imaging team within the OPPO research and development (R&D). We are in charge of the development and optimization of “Ultra Night Mode" and “Ultra Dark Mode”, as well as the enhancement of HDR (high dynamic range) experience across Reno Series and FIND Series. (Editor's note: the Reno and FIND comprise their mid-range to top-of-the-line devices)
We are committed to responding to users’ demand on photography, and continuously upgrading the technology to fit their user habit. We have 2 major objectives: first, to continuously enhance the technology; and second, to be user-centric and consider the users’ point of view to be able to address their pain points.
We need to keep broadening the photography scenarios in order meet more of the users’ needs.
Q: What makes “Ultra Night Mode” and “Ultra Dark Mode” unique compared to other night modes in other phones?
WL: Our aim was to enable users to break the restrictions of light when taking nighttime photos.
It allows users to take clear shots in dark environments, even in extreme scenarios where there is almost no light.
The competitive advantage of our low-light photography is that not only are we able to capture dynamic range of night scenarios, but we have also shortened the development time of the image to up to 4 to 6 seconds.
For the front camera Ultra Night Selfie mode, we’ve addressed the pain point of taking selfies at night by bringing stunning night [selfies]. The algorithm has already been used in the rear camera, and it's our first time applying this to the front camera [on the Reno3]. The advantage of ours is that we have a special optimization that supports face protection. This protects the face’s brightness and clarity through face detection, and it takes 5 seconds to develop the image.
Q: Why do you think that low-light photography is the current big trend now in smartphones?
WL: We’ve always put immense efforts to study what consumers are looking for in their smartphone. The pain point of dim-light photography being not clear, and bright enough has always been an issue.
To demonstrate that we’re strong in photography, night mode is one of the four major indicators for imaging performance. In other words, we believe that a smartphone that is able to take clear photos in dim light equals a good camera phone. Since the R17 Pro, we’ve been doing different optimizations on the feature to make sure night shots get brighter and clearer. OPPO has also further optimized the frequently used night scenarios such as being under a streetlight or in an indoor speakeasy bar.
Q: Aside from low-light photography, what areas in phone photography need to be improved next?
WL: Photography has always been a major focus of technology R&D at OPPO. We particularly pride ourselves in our zoom technology. I believe that going forward, we’re going to continue enhancing the foundation of photography, in areas such as achieving shorter image development time, reducing noise in the photo, and improving the stability of video shooting.
We've also greatly improved on the accuracy of scene recognition through a large amount of image materials training. In the future, we want to further optimize the accuracy of scene understanding through AI, enabling users to take more authentic photos anytime, anywhere.
We're also seeing that video recording demands are growing rapidly, alongside the photography function that has been growing exponentially over the past few years.
There is increasing demand on visual expression from young users, and the demand for video creation. In China alone, there are 249 million vlog viewers in 2019, according to iiMedia Research, while worldwide, there are over 4 billion viewers watching video content every month, according to GlobalWebIndex.
With the advent of 5G, there'll certainly be new camera upgrades that'll come with all that extra bandwidth, say, perhaps, real-time 3D video conferencing capabilities.
We're aiming for the Reno series to grow with customers' need in photography, video, and entertainment. In the Reno3, in particular, we're proud of its steadier video, video bokeh, and video zoom features.
Q: Phone cameras are so great now but you still know if a photo is taken with a true DSLR or mirrorless. Do you think that we’ll ever reach a point where the quality will be indistinguishable? How long do you think it will take before we reach that?
WL: Our camera R&D team intends to push the shooting performance in various scenarios – including night scene, wide-angle, telephoto and video – certainly to a quality that is comparable to a DSLR.
Such higher standards bring unprecedented technical challenges. But it's something we have decided to approach holistically, particularly during the development of the OPPO Find X2. (Editor's note: the Find series is a line that showcases OPPO's latest tech.)
Our process is we’ve reviewed the development of the mobile phone camera, looked for gaps by comparing photos shot with mobile phone cameras and professional DSLR cameras, and re-examined the five underlying sensor specifications, namely: photo-sensitivity, dynamic range, color depth, focus performance, and signal-to-noise ratio. We can certainly say that the X2 has seen considerable improvements in these areas, coming from our efforts last year.
It's a technical challenge, but as an engineer, we're always looking to move closer to these higher standards. – Rappler.com