Công nghệ

Computing photography, users benefit

In India, the second largest phone market in the world (after China), The Times of India newspaper said that the camera is a very important factor for users when considering buying a smartphone. Therefore, phone manufacturers are racing to improve the photography function on their products.

Using computational algorithms to reduce costs

The mobile photography function is based on both hardware and software, so if you want to optimize it, you have to combine both parts.

However, in reality, the investment in hardware is very expensive, causing product prices to rise, reducing competition in the market. In addition, camera hardware is mainly provided by 3rd parties, leading to manufacturer dependence. Meanwhile, software depends on the manufacturer’s own internal resources, which is a place for manufacturers to show off their own technology, creating a different mark for the product.

Photography algorithms have evolved into computational photography (NADT), which uses computational algorithms to process images regardless of the tiny size of lenses and sensors on mobile devices. Accordingly, improving image quality beyond what lenses and image sensors can record, reducing motion blur, adding more depth of field, improving color, contrast and light strip. With artificial intelligence (AI) technology, NADT is like a dragon with wings on the clouds! This is the reason why the Apple family does not fall into the race to equip the increasingly “terrible” camera clusters of the Android world, but still persists in focusing on improving the photography algorithm for its iPhone lines.

On the flagship iPhone 13 Pro Max launched in September 2021, Apple still only uses a cluster of 3 12 MP cameras. Even in the Android world, Google – the owner of the Android operating system – has pioneered the development and application of NADT. And NADT became a hot topic when Google launched its Pixel 2 smartphone in 2017. On this device, Google used Machine Learning to process multiple images taken with the same location, objects in different locations. different angles to create a larger, more detailed scene. And until the Pixel 6a generation was announced on May 11, while other companies have equipped with 4 cameras up to 108 MP, Google is still “calmly” using a configuration of 2 12 MP cameras. Until now, mobile photographers still consider the Pixel line with NADT as one of the “masters of photography”.

Computing photography, users benefit - Photo 1.

OPPO’s MariSilicon X dedicated imaging NPU chip

The future of camera technology

In Vietnam, Bkav from the very beginning has been persistent in applying NADT for its Bphone smartphone. And to be fair, the Bphone line has produced shimmering images.

Even Samsung, the world’s No. 1 smartphone manufacturer, has persevered for many years to stay out of the race to upgrade the “terrible” camera, but still uses the highest configuration, the 12 MP camera. It wasn’t until the Galaxy S22 Ultra and Galaxy Note20 Ultra series in 2020, that Samsung followed the trend of using a 108 MP camera. Even though it has now used 64 MP and 108 MP cameras, Samsung still emphasizes on AI applications. During the launch event of the flagship smartphone OPPO Find X5 Pro in Ho Chi Minh City on the evening of May 5, Mr. Do Quang Kha, CEO of OPPO Vietnam, said that OPPO has applied NADT for this flagship line. “NADT will be the future of mobile camera technology in the future” – Mr. Kha said.

OPPO, the world’s 4th largest smartphone manufacturer in the first quarter of 2022, played big in the smartphone race. They have researched and developed the world’s first image processing AI (NPU) chip called MariSilicon X and introduced it at the OPPO INNO DAY 2021 annual event on December 14, 2021. OPPO Find X5 series is the first smartphone series equipped with Imaging NPU. According to OPPO, this dedicated imaging NPU delivers impressive performance with optimal power efficiency, and is capable of real-time RAW image format processing. The processing power of this dedicated NPU enables incredibly high-quality 4K AI Night Video live recording and preview – a challenge for today’s mobile cameras.

Camera lenses, no matter how advanced and advanced, are just inanimate components. Only the NADT can recognize the type of subject (people, food, pets…) for appropriate handling. AI will consult a worldwide database to synthesize and adjust for optimized images that please the photographer. Of course, after a period of “learning” and improving, AI in photography is becoming more and more natural, more real, less magical, such as processing human skin color.

Camera hardware is considered the heart of a smartphone and it also accounts for a large part of the device’s cost. TechInsights said that in 2017, the share of camera hardware in total smartphone material costs was 12% (with HUAWEI Mate 10), 8% (Samsung Galaxy S8) and 7% (Apple iPhone 8). According to The Korea Herald, a cluster of 3 cameras, the highest of which is the 108 MP main camera, on the Samsung Galaxy S20 Ultra in 2020 is worth up to $ 89.17. Meanwhile, the 25MP camera on the Samsung Galaxy A50 costs $19.3. Cluster of 3 12 MP cameras on Apple iPhone 11 Pro Max also costs up to 42.4 USD. Apple also had to pay $34.5 for the camera hardware on the iPhone XS Max. And Huawei had to spend $ 56 for the camera cluster on the P30 Pro smartphone. The unnecessarily high pixel count definitely increases the cost of the device.

Mobile users, especially those who love taking photos on mobile devices, will certainly benefit from NADT. The race in photography functions on smartphones is now actually a race between the photography algorithms of mobile device manufacturers.

Increase image quality

Camera hardware has its limits, while NADT has almost limitless possibilities. Due to the specificity, smartphones cannot accommodate full-sized image sensors and professional lenses. It is the versatility of the NADT that will help the smartphone photography function to overcome those hardware limitations. Image quality also depends on the pixel size, which the larger the pixel, the more light and image data it can capture.

You are reading the article Computing photography, users benefit
at Blogtuan.info – Source: nld.com.vn – Read the original article here

Back to top button