Data show that, since 2015, the state has made a significant effort to boost net reductions, with more than 95 per cent of mobile-flow charges and more than 70 per cent of corporate broadband and line charges。
At the same time, the age of data explosions has come. Data from the national survey of data resources (2023) show that in 2023, total national data production amounted to 32. 85 bytes (zb), an increase of 22. 44 per cent over the same period, mainly due to the explosion of non-structured data such as content creation, video and video。
In financial terms, however, the absolute value of bandwidth costs has declined. The data show that in the second quarter of 2023, the bandwidth cost of the achilles was 450 million, compared to 2. 4 billion yuan spent throughout the year on achiques in 2020, an average of 600 million yuan per quarter。
In an interview with the it times journalist, ppio-ceo yao xin stated that the cost of operating the video site remained extremely high due to the explosive growth in user demand for high-resolution video。
Video platforms therefore need to strike a balance between controlling costs and enhancing user experience, for example by introducing more advanced video coding techniques or using distributed programmes such as marginal cdns to reduce high costs。
Should we compress the video? It's inevitable, but it shouldn't be reduced
In industry's view, compressed video on video platforms is a necessary option for “optimizing cost structures”。
Yao xin was a pioneer in the video age of the internet. In 2004, he dropped out of school at the chinese university of science and technology, invented the p2p-streaming protocol and created a video platform pptv, covering 450 million users worldwide, one year before youtube。
He pointed out to the it times journalist that it was an ongoing process of optimization that businesses had to strike a balance between cost and user experience. In order to reduce bandwidth and storage costs while balancing user experience, video sites had to be coded with high compression rates。
Over the past two decades, video-coding technologies have evolved from h. 263 to h. 264, to the present time h. 265, with each improvement in coding technology leading to a 30 to 50 per cent drop in the number of code flows, from 720 p to 4k and 8k. As tim explained in hurricane video explains, because high-resolution videos (e. G. 4k) occupy huge storage space, video data can be as high as 44gb for one minute. High compression coding has significantly improved video quality and effectively controlled costs。
Video platforms abroad, such as netflix and youtube, are equipped with specialized algorithm teams to optimize video transmission. For example, more bandwidth resources are allocated to periods of intense change in the picture by reducing the number of codes at times when the picture is less variable, thus enhancing the viewing experience of users even when the resolution remains unchanged。
In march 2020, disney announced a reduction in video bit rates across europe to help control internet traffic during the epidemic, as did netflix, youtube and amazon platforms. In november 2023, sony announced the acquisition of isize, a british company focused on in-depth study of video transmissions, to use ai to improve the visual quality of video stream content and reduce the amount of data required for serial video games。
But why is it that, after the fermentation of the incident, many people feel that the 1080-p video that was viewed on the national video site is less clear than the 720-p video platform outside the country
In fact, while the country has uniform broadcasting standards for digital television, the industry standards for video sites vary. The so-called “high-clean” (hd) and “super-clear” definitions differ。
This technological innovation depends on the capabilities and technological strength of each platform. As a result, the nominal resolution of different video sites may vary. If there is insufficient technical strength, the nominal 2k of some platforms may have the same effect as the 1080p of others。
In addition, yao xin believes that different types of video content require different quality requirements, such as animations, sporting events and visual works, with different standards of quality and code rate. Video sites must therefore be weighed against the type of content and user needs。
In any case, it is the principle that the global video industry should follow to improve coding efficiency and achieve better quality at the same flow rates, even at lower levels。
The reporter's handwritten book, "technology, not deception."
A survey of journalists found that the decline in the quality of video was a complex issue of multiple interests behind which “data pressure” had become one of the major risks facing modern society。
Although the current perception of a change in the quality of the painting may not be apparent to ordinary users, the uninformed “replacement” or downscaling behaviour of the video platform is undoubtedly an infringement of the right to know and to fair trade for creators who, like hurricane “film” and consumers who pay for high-quality videos. Individual video platforms use labels such as “high-coded” “no-lost” as a means of attracting users, but in practice they conceal the true situation through technical means and are contrary to good faith。
As yao xin has said, the core source of revenue for video sites remains the realization of advertising and user fees, which are extremely expensive to access. Therefore, the platform should not easily sacrifice the user experience in its decision-making, which, although the bandwidth cost is still not low, is secondary to the cost of acquiring users。
In fact, with the development of artificial intelligence-generated content (aigc), the production process of video will be subverted and the cost of production will be further reduced, which will boost the spread of more clean video。
Today, artificial intelligence is deeply embedded in video production and transmission. Using an in-depth learning model, content sensor coding can be achieved to adjust compression rates to the dynamics of video content. For static or less variable scenarios, a higher compression rate is used, while for frequent or detailed scenarios, lower compression rates are used to preserve key details. In addition, ai technology can be used for ultra-resolution reconstruction of videos, real-time enhancement of user-end video clarity and reduced reliance on high bandwidth。
New technology applications in the field of transmission, such as the marginal cdn provided by the ppio-euyun, also show important potential. Marginal cdns can significantly reduce delays and increase transmission efficiency, optimizing the viewing experience, especially in time-sensitive scenarios such as live broadcasts and high-spectrum videos, by slowing audio-visual content to the nearest edge of the user。
At the same time, advances in the new generation of video coding technologies have provided new ideas for improving user experience. For example, h. 266/vvc (multipurpose video code) can reduce data by about 50 per cent while maintaining the same video quality. These advanced coding techniques effectively compress video data through more sophisticated algorithms while retaining the maximum amount of video detail。
As technology evolves, a higher bandwidth and less delayed network will provide better transmission conditions for ultra-high-resolution videos, while the spread of more efficient coding formats and transmission protocols will also enhance overall transmission efficiency and user experience。
The challenge for medium- and long-sized video sites to achieve efficiency gains in reducing bandwidth and storage pressures without sacrificing user experience will be an increasing number of “down-end” paths。
Or the old saying: technology is first productivity。
Layout/ jigyan
Photo / b site network it times
Source/it times public name
E n d
