Photography is one of the most essential topics when it comes to mobile AI. No wonder several new smartphones – such as HUAWEI P20 Series, Xiaomi Mi MIX 2S and OPPO R15 – feature AI photography.
As AI becomes more popular, AI capabilities and cloud computing have grown exponentially, and speed up graphics processing, the question that pops up in one’s mind, why is mobile photography associated with AI?
However some people remain skeptical of AI – they see mobile AI photography as a gimmick enabled by graphics software. That’s indeed a good thing of a working brain that it can think.
But the recent launch of HUAWEI P20 Series has changed their perception. The phone won outstanding reviews from DxOMark, a leading source of independent image quality measurements and ratings for smartphone camera and lens. It has the hardware that is dedicated to work for photography (e.g. photoreceptors and triple camera setup) and software such as Predictive Focus. If we take a step back, we can see the magic between AI and mobile photography.
Beauty is greatly appreciated, hence the birth of AI photography
Smartphone users are not satisfied with their pictures these days, which is nothing new.
Despite the progress from the 5MP camera on “noble” Nokia, to 20MP HD shooters on other common models, phones failed to produce an Instagram-ready selfie. People needed clear and stunning photos, be it landscape or portrait.
For a while, phone makers had been struggling to delight their customers. While they were busy improving the pixels and zoom levels, tons of photography apps emerged to meet different needs – from portrait-based Meitu to Snapseed for post-processing. There was even an app that created double exposure effects.
Back then, we were closing in on a new era of AI photography where smartphone users weren’t keen on picture apps any more. Instead, they were looking for something more intelligent: a beauty app that turned on automatically, or detected scenarios and added filters to get the perfect shot. That said, it was ok for them to use different apps, point and shoot, and upload pictures to the cloud for deep processing.
Mobile phone manufacturers soon realized they needed a solution to overcome the constraints of limited computing power. But facial recognition, neural style transfer, and other tech designed for more beautiful portrait and image processing only work on the cloud at the expense of user experience.
By improving the computing power of the device, AI allows scenario identification and real-time image processing to enhance photography experience, even if the pixel levels remain the same. That’s where the widespread use of mobile AI photography comes in.
Sensors and computing power for capturing beautiful moments
When it comes to AI photography, we should never underestimate the importance of AI capabilities that boost computing power. Better computer power is a must for smart power saving and resource allocation on all AI-featured smartphones such as HUAWEI Mate 10 Series, HUAWEI P20 Series and HUAWEI Honor V10.
The same is true for mobile AI photography.
Better hardware is widely seen as a primary reason behind HUAWEI P20 Series’ DxOMark reign. The 1/1.7-inch sensor has better light sensing capability and shows excellent details.
But it is the outstanding computing performance of the AI that enables the powerful hardware on the HUAWEI P20 Series.
That’s why we have seen the latest breakthrough devices with better cameras. They should run fast enough for users to get a perfect shot and enjoy greater photography experience.
A good case in point is iPhone X, which positions portrait lighting as their key camera feature. With simulation algorithm, the camera integrates studio light effects with the face of the subject. This feature is closely related to the 3D light sensor of the front camera because the capability to identify the facial structure and details is essential for this effect. The bottom line is: The sensor requires massive computing.
Another example is Google Pixel 2, which won 98 scores from DxOmark, beating iPhone X at 97. What’s interesting though is that while most phones feature dual camera, Pixel 2 manages to achieve better photography with just one camera. How? Google integrates an IPU (Image Processing Unit) and many sensors into the camera, to collect and analyze spatial and depth data.
HUAWEI P20 Series has taken the integration of AI performance and camera hardware to the next level.
Take the Super Night Mode as an example. The performance of the light sensor determines the picture quality in low light environments. A bigger sensor means larger pixels, which increases light sensitivity and captures more details.
However, even with excellent light sensing capability, it is still difficult to produce night shots as fast (e.g. 1/30s or even 1/500s) as one could during the day. Instead, users may need 2 seconds or more to expose the sensor to enough light. Any motion such as camera shake may ruin photos with high noise levels.
The Super Night Mode adopts a motion recovery algorithm which aligns all the images during the prolonged exposure to create clear pictures. The imaging process is completed within just seconds. Without the powerful computing performance of the AI, it would be impossible to finish the high-quality imaging and algorithm processing quickly.
The device also features predictive focus for moving subjects and 3x telephoto lens for long-range photography, creating better pictures easily. They are all powered by the AI that uses AI to process millions of images all the time.
An Alternative to DSLR Camera
With HUAWEI P20 opening up a new world of mobile AI photography, many people start to wonder: What’s next? One strong voice says smartphone photography will be stronger and eventually replace DSLR cameras.
It’s easy to see why people may think that way when HUAWEI P20 is launched. Its light sensor is as good as that of a compact or digital camera, and delivers DSLR RAW format to preserve more details for post-processing. There are other features such as the professional mode similar to the M setting on DSLR camera. “M” means manual adjustment of focal length, aperture, and shutter.
The truth is, despite all these features, smartphones can hardly “replace” DSLRs in terms of hardware. The flexible combination of mechanical shutter, multiple lenses, and flashes on the DSLR is impossible on mobile phones. Moreover, users will be overwhelmed by the basics about DSLR such as white balance, aperture and ISO values, let alone how to frame or edit pictures.
To some, the only thing they know about DSLR is to press the shutter. Ironically, the renowned Chinese director Peter Chan used iPhone to make a short film called “Three Minutes”. This shows that the secret to excellent photography does not lie in the hardware of a DSLR and smartphone, but it’s the filming experience of a pro and amateur.
With AI technology, smartphone users can produce DSLR-like images without learning any complex photography knowledge. Thanks to scenario identification, portrait lighting and Super Night Mode, users don’t have to adjust parameters or use a tripod.
Instead of replacing DSLR, mobile AI photography offers another possibility to record their beautiful memories even with little or no photography knowledge.
As its name suggests, Master AI on the HUAWEI P20 Series means empowering ordinary people with the experience and ability of a “master” photographer through AI. That is what AI means for Huawei.
Mobile AI photography cannot realize its full potential without advanced camera hardware and software. Moving forward, when we shift our focus to the horizon of light and shadow, we hope everyone can capture their beautiful moment like a Pro.
