Every so often a new technology comes along that serves as a lightning rod, reigniting conversations around the impact of technology on jobs, culture and our way of life.
These days, that technology is Artificial Intelligence (AI).
From people praising the potential of AI to automate and streamline, to others calling for a six-month moratorium on large-scale AI research, it’s become one of most polarising topics in ages.
Despite this, one topic that has received very little coverage — other than a recent Samsung snafu where it was discovered that the brand was using AI to “enhance” and add detail to images of the moon — there hasn’t been much talk about AI in smartphone photography.
“The best camera is the one that’s with you,” is a quote that many of us have heard and may even have used from time to time. These days that camera is usually the one on your phone.
Camera quality and capabilities have become so important that four years ago CyberMedia Research (CMR) released a report stating that “89% of buyers consider the camera as the top specification while buying a smartphone.”
If we look at social media, it’s clear that platforms that prioritise visual mediums, including photography and videography, are continuing to rise in popularity versus their counterparts that don’t.
The ability to quickly capture an image or a short video has also resulted in more people wanting to create content.
Speaking about the images submitted for the HUAWEI Next Image Awards, Dr Nichole Fernandez, a visual sociologist with a PhD from the University of Edinburgh, said: “The images show that these camera phones are powerful, turning everyone into not just a consumer of images but a creator. Even the untrained person can become a photographer, capturing moments they find meaningful and creating something they are proud enough of to submit to a contest. It is democratising image production.”
None of this would’ve been possible without computational photography, or as some brands like to call, the AI camera, which uses software algorithms to process, and in some cases, enhance the images you take with your phone.
“Since the first commercial camera phone was released in Japan in May 1999, this unique mobile device feature has been on a continuous journey of multiple upgrades, from multiple lenses, optical zoom, and image stabilisation, right up to today’s use of sophisticated software products to capture those special moments.
MediaTek Imagiq transforms the experience behind the cameras by developing industry-leading multicore, HDR-ISP designs that provide extremely high performance with exceptional power efficiency.
The unique HDR technology works by combining dedicated HDR engines to groups of ISPs, enabling MediaTek’s smartphone SoC’s to capture three different exposures per frame for photography, or while reliably maintaining 30 fps recording during videography.
Today smartphone users expect faster, more accurate exposure, focus, white balance, and noise reduction, even with HDR capture. However, what many are unaware of, is how their smartphone works “behind the scenes,” to take the best possible photos. MediaTek’s 6th Generation MediaTek APU, an AI for smartphones — enhances HDR capture with a range of post-processing optimisations, providing the most stunning photography and videography results. Advances in smartphone hardware include faster and larger resolution displays, increasingly large and complex camera systems, faster always-on connectivity, larger caches, bigger storage, more sensors and features than ever,” says Rami Osman, director for corporate sales and marketing at MediaTek Middle East and Africa.
Addressing some of the issues plaguing early smartphone photography were companies like Huawei who were, and still are, one of the leaders in smartphone photography.
When Apple unveiled its neural processing unit (NPU) in 2017, Huawei unveiled its own NPU barely one month later.
Unlike other brands that released NPUs, Huawei’s processing was done on device and was able to help machine learning and AI optimise for various things including photography.
The fact that you can simply point your phone at something, snap a picture and the camera is able to recognise what it needs to capture and accurately reproduce all the details from colour to sharpness, focus and adapt for lighting, is because of the NPU.
Over the years, these machine learning and AI algorithms have got smarter, more powerful and more capable, resulting in night-time and lowlight photography that boggles the mind, as well as the ability to take pictures using 100x zoom.
Our cameras now also help us recognise objects, do visual language translations and so much more.
While the focus has predominantly been on software, technology brands shouldn’t forget the importance of hardware and should be looking at ways to find the best marriage between the two.
Speaking about the Huawei Mate 50 series and the P60 series, both of which sport the brands new XMAGE capabilities, Akhram Mohamed, vice-president, operations at Huawei Consumer Business Group, said: “The software side is complemented by the hardware, the sensors, the mechanics of the actual lenses and everything working in tandem. Each time we produce a product, we try to improve it just a little. If you look at the P60 series, you’re talking about different scenes, filters, lighting effects and the amount of light that is taken in. One phone might be better in lowlight photography and another might be better in zoom capability. The underlying aspect of Huawei's cameras is that we’re not going to enhance just one area, whether it’s computational or hardware, we’re going to put them together and see how we can help improve photography in general for our users.”
Talking about the recent spate of smartphone brands partnering with well-known photography brands and in lieu of investing in and developing their own imaging capabilities, Mohamed said: “It’s not that this Huawei (the P60 series) isn’t doing it any more (using an in-house made NPU — something the brand is no longer able to do as a result of US restrictions) and we had to look at alternatives to do it, it’s that no-one else is really focusing on it either. Everyone’s doing computational photography but it’s a very generic way of doing it. When everyone has the same, there’s no more innovation.”
The question to ask when it comes to these smartphone and photography brand partnerships is whether there’s any real benefit or if it’s just a marketing strategy and what these partnerships, and the overall evolution of smartphone photography means for traditional camera brands.
David Imel, lead researcher and writer at MKBHD and Waveform podcast host, says: “Computational photography techniques in smartphones are still advancing faster than traditional cameras and sensor technologies, and that’s because traditional cameras are physics aided by processors, vs smartphone cameras which are primarily processors pushing the physics. Processing speed and capabilities on average do still advance faster than we find new hardware techniques.”
“With this said, I have seen a noticeable change in how many smartphone companies are handling larger sensors. Companies like Xiaomi in particular are starting to take advantage of a larger sensor’s natural characteristics and pulling back on the computational enhancements of those sensors, which leads to images that look much more like that of a dedicated camera, with harsh shadows, bright highlights, and less of a “plastic” look,” Imel said.
“Some companies still process too intensely, even with larger sensor technology. Most of the positive rollbacks are coming from Chinese smartphone manufacturers, and it would be nice if other global brands followed their lead.”
It’s clear that smartphone photography, and therefore AI and machine learning aided photography, is here to stay and while that may not be a bad thing, we need to consider just how much impact it will have on society when the very images you capture are not a true reflection of reality.