At the beginning of the 14th century, Giotto Dibondone creatively integrated the knowledge of architecture, mathematics and physics into the creation of frescoes, thus giving birth to the Florentine School. Under the foreshadowing of the latest science and technology, the techniques of this school quickly spread throughout the Apennine Peninsula, creating a real artistic peak: the Renaissance.
Renaissance art may be a model of aesthetic and artistic creation based on science. Since then, human beings have proven time and time again that at the top of science and technology, it is often possible to find art that makes ordinary people indulge in it – for example, mobile phone images.
Today, we can easily take a photo that fascinates us with Huawei mobile phones without requiring professional photography skills or complex photo editing. We might even forget that there was a time when cell phone photography wasn’t so good.
In 2000, Sharp launched the world’s first mobile phone with a camera, the J-SH04, a CCD camera with only 110,000 pixels. For a long time after that, the upgrade of mobile phone photography mainly focused on pixel improvement. It is always difficult to use mobile phones to overcome many challenges of light, distance, and color, and to take satisfactory photos.
It may be said that from mobile phone photography to mobile image art, we always have a threshold that we have never crossed.
But in recent years, it should not be difficult for you to find that the imaging capabilities of mobile phones have been greatly improved. Maybe it’s an AI photography assistance; maybe it’s a shooting in the dark; maybe it’s a super long-distance capture; or maybe it’s a momentary snapshot of high-speed motion – today’s mobile phone imaging capabilities in our hands are no longer a pile of pixels. It is the symphony of intelligence and computing, the concerto of dark light, backlighting, telephoto, and macro – mobile phone images begin to understand people, making every user a master photographer.
Today, every mobile phone that wants to hit the high-end market must come up with a complete set of imaging solutions; every mainstream mobile processor must work hard on AI computing power and imaging capabilities; every major mobile phone manufacturer is planning and the new possibilities of computational photography.
Looking back, maybe we are suddenly a little confused, how did this change come about?
In 2017, we first predicted that Huawei would deploy AI capabilities in mobile chips to solve targeted computing problems in vision and imaging. In the past four years, mobile phone images have changed far more than we expected at the time. There may only be one explanation for this: Huawei’s changes in AI photography and mobile imaging have eventually grown into an irreversible and profound change in the industry.
In this changing situation, the complexity of the imaging capabilities of mobile phones has been comprehensively improved; the relationship between computing, intelligence, and lenses has been completely reconstructed; while the technical complexity has been upgraded, the experience threshold of mobile imaging has been greatly reduced, allowing the public to experience The beauty of images.
Driven by technical collaboration, end-to-end intelligence, computing systems, and multi-camera collaboration, these completely unfamiliar words a few years ago have joined the “renaissance” of mobile phone imaging. Promote the image from the pixel era to the intelligent era with vigor and vigor. And Huawei’s industrial influence in the mobile phone market constitutes a well-deserved signpost in the entire change.
Before the mobile video feast in 2021, let’s go back to the whole story. Let’s relive this renaissance of mobile phone images together. Turning mobile phone lenses into art, the story begins in the fall of 2017.
Today, AI technology has covered streets and alleys, and face recognition has become everyday life. But in 2017, all walks of life are still in the discussion of the use of AI technology, and fewer people combine AI technology with mobile photography. Even if some large technology companies have proposed applications related to AI image optimization, they do not think that AI, as a software algorithm, must be supported by independent hardware units.
However, at this point in time, Huawei accurately realized that AI can become a key to mobile phone images. Looking back at its logic, Huawei’s starting point for insight is that at that time, the imaging capabilities of mobile phones had become more professional and the parameters continued to be refined. But the mobile phone is obviously not a SLR, and the beauty of its images should be within reach, rather than requiring long-term skill learning and technical polishing. In order to achieve a balance between the technical complexity of mobile phone images and the tendency of user experience to be simple, AI’s own recognition, judgment and scene-based output capabilities seem to be the best answer.
On the other hand, many pain points in mobile phone images can be solved by AI. For example, the anti-shake ability in night shooting mode can be enhanced by AI technology, and the predictive analysis ability of AI can be applied to motion capture and video shooting. In order to achieve a comprehensive improvement in the imaging capabilities of mobile phones, it is the general trend to fully embrace intelligence.
The next insight is that Huawei found that AI processing capabilities must be placed on the device side. On the one hand, this is because the image data in the mobile phone belongs to the high degree of privacy of the user, and the completion of the processing by the AI computing power of the cloud always leaves hidden dangers of privacy; on the other hand, the beauty of the mobile phone image should be real-time and immediate feedback. If users have to wait overnight for a better photo, that’s obviously going to hurt the landscape.
Put user value at the core and exclude all other options, then even if the rest of the road has never been traveled, it must be correct.
The end result is that in the fall of 2017, Huawei launched the world’s first mobile SoC chip Kirin 970 with a dedicated AI processing unit NPU.
Subsequently, the Mate10 series equipped with Kirin 970 launched the AI photography concept for the first time, officially introducing end-to-end AI into the mobile phone industry.
In 2018, Huawei continued to dig deep into the chemical reaction between AI and images in the P20 series, and launched the “AI Image Master” function for the first time. The P20 series has a unique intelligent mode recommendation, which can automatically adjust and set parameters for different scenarios, and can be used in real time. It locates the subject of the photo, automatically recognizes the attributes of the object, and presents different algorithms for objects with different attributes. It has a total of 19 categories and more than 500 recommended photography scenes. In the handheld night scene mode of the P20 that amazed the entire industry at that time, it also integrated Huawei’s AI anti-shake technology and AI algorithm recognition technology allow users to give up the tripod and still be able to capture the gorgeous night.
In the following years, we can see that almost no terminal manufacturers do not launch AI photography capabilities, realizing the grand situation of “no AI photography without mobile phones” at home and abroad. Qualcomm and Apple have also followed Huawei’s judgment on the AI processing capabilities on the device side. The AI task processing unit has become an essential element of mobile phone computing after CPU and GPU.
In those days, Huawei found the key to AI, and the mobile phone image also found the key to Huawei.
Taming time and space
In the following years, we have officially witnessed the explosive development of mobile phone imaging. In 2015-2016, the homogeneous competition in the mobile phone industry was the most widely discussed topic in the industry. But today there are very few similar voices. There may be many factors, but it must include Huawei’s proof in the field of mobile phone imaging that the industry is far from reaching the ceiling.
To sum up Huawei’s breakthroughs in mobile phone imaging over the past few years, we may be able to use the word: tame time and space.
For mobile devices, the natural limitations of weight, light input, and camera carrying space make it difficult for image creation to overcome the interference from time and space. The change of time brings the sun and the moon back and forth, and brings us problems such as dark light, low light, and backlighting;
And Huawei’s image breakthrough is to continuously complete a breakthrough about time and space. From dual-camera, triple-camera, quad-camera, and five-camera, from Huawei P9’s dual-lens Leica color, to P10’s portrait photography, from P20’s dark-light night scene mode, from P30’s periscope zoom architecture, to P40’s full-scale camera Time period, all focal length ultra-clear images. We have witnessed the Leica color that the P9 began to make users addicted to; in the P20 series, we encountered a 1/1.73-inch super-large image sensor that was extremely rare in the industry at that time; in the P30 series, we sensed the unruly and imaginative RYYB spectrum; in the P40 Pro+ , ushered in an extremely shocking 10x optical zoom and 100x binocular zoom periscope camera that far exceeds the limit of the human eye. Along the way, the night is tamed, the colors are renewed, and even the limits of light seem to be broken. Users will be curious, what else is there that mobile imaging can’t do? At the same time, he is extremely looking forward to Huawei’s next extreme challenge on mobile phone images.
At the same time, we see that more mobile phone manufacturers choose to cooperate with professional optics and imaging companies, and technologies such as multi-camera, ultra-wide-angle, telephoto, and dark shooting have repeatedly become the imaging road led by Huawei and the common development of the industry. Huawei’s continuous iteration and innovation in imaging has not only given Huawei a firm voice in the high-end market, but also allowed more manufacturers to find a way to go high-end.
The ultimate answer to taming light and shadow is that the mobile phone industry has healed the heart that used to be skeptical about the future.
From the rise of intelligence to the continuous iteration of software and hardware, when the time comes to 2020, the accumulation of past images begins to point to the same new direction: computational photography.
In 2010, Stanford University held a technical salon, the theme of which was to discuss whether it is possible to let the photographic equipment program the images captured by the computer like a computer. In this way, photography is no longer the world of photographers and refiners, but is open to all ordinary people. But the result of the discussion at the time was that all SLRs, card machines and smartphones could not meet this need.
The core idea of the so-called computational photography is that mobile photography is not just a mobile phone + a camera. Instead, I hope to mobilize the computing power, algorithm ability, and hardware coordination ability of the mobile phone when shooting, and integrate the intelligence of the computer with the ability of the camera. This can maximize the computing and processing capabilities of mobile phones that are different from professional cameras, and can also lower the threshold for users to achieve the integration of intelligence and art.
In order to achieve the goal of computational photography, on the one hand, mobile phones need to have strong AI computing power and image processing capabilities; on the other hand, various software and hardware systems of mobile phones are required to achieve a high degree of integration and collaborative control. It may be said that the accumulation of computing power, imaging capabilities, and intelligence capabilities in the mobile phone industry over the past few years is preparing for the step of computational photography.
In the Huawei P40 series, the XD fusion solution that combines the previous AI + image processing capabilities in Dacheng was launched, announcing that Huawei became the first batch of players to enter the computational photography track. We have summarized the three main capabilities of XD fusion”
1. Coordinate multiple shots
How to coordinate the image acquisition of multiple cameras is the first task of XD Fusion. For every additional camera, there is an additional data input source and a series of superimposed task processing schemes. At this time, the XD Fusion needs to integrate the main camera with the wide-angle, TOF lens, telephoto lens, and even the fusion between the P40 Pro + binocular telephoto, so that the multi-camera system is more intelligent than the lens combination.
2. Scene recognition and processing
The second function of XD Fusion is to segment, understand and process image semantics based on machine vision technology. Through high-level semantic analysis, let the mobile phone know what the object is, so that targeted optimization can be carried out, especially to ensure the implementation of AI capabilities such as removing passers-by and removing reflections.
3. Image quality upgrade
In the computational photography system, each photo contains issues such as signal-to-noise ratio, full well capacity and so on. The ability of XD Fusion is that it handles image quality, color, light and shadow, dynamic range, blurring, etc. through a neural network system, making the final image instantly refined.
And for users, we probably don’t need to understand what the XD Fusion looks like when it’s disassembled. Just know that mobile phones can bring a steady stream of video surprises. Huawei often says that it keeps the complexity for itself and the simplicity for users, which has been best confirmed in the era of computational photography.
Today, Apple and Google are already in the field of computational photography, and it may take time for the Android camp to join. But Huawei’s industrial layout undoubtedly constitutes a clear signpost. We will wonder how the age of computational photography will come, but we will never worry about whether it will come true.
Conclusion: The road sign of the industry, the beautiful field
After several years of development, AI, multi-shot, night shot, super telephoto, computational photography, these cool-sounding words have become our friends when taking pictures. A wonderful feeling is that when I was asked by others to take pictures at the scenic spot, I felt that I didn’t know how to start when I picked up someone else’s mobile phone. Maybe intelligence and computing are like that, they come to you silently, and after a while you find that you can’t leave.
Predicting the next stage of imaging breakthroughs, we can assert on Huawei’s imaging development road that there is still room for improvement in the software and hardware collaboration capabilities of mobile phones. Mobile imaging will have a higher degree of integration, and computational photography will be unshakable in the future. development direction. The root of this judgment is that Huawei has a clear industrial roadmap in the imaging field, and has continued to invest in large-scale R&D in the identified value direction, and has continuously upgraded and optimized the technology system after anchoring. This cycle of continuous integration and re-breakthrough turns images into a long-term track with sustainable growth and abundant hardware and software bases.
The entire mobile phone industry has benefited from this, and has an industrial space for long-term upgrades, investment, and healthy competition. For the technology industry, competition and catching up are not terrible, but the real harm is homogeneity and lack of innovative direction.
For mobile phone manufacturers, computational photography and mobile imaging are the same road; but for users, it is a field about art, creation, and beauty.
If the Renaissance allowed human art to have unlimited imagination. In the computing and intelligence era of mobile phone images, road signs also point to infinite possibilities.
Pick up your phone and you are Da Vinci.