Android IOS WebRTC Audio and Video development Summary (48)--the opportunity to see the video industry from a commercial and technical perspective

This article mainly from different angles to introduce the opportunity of the video industry, the article from the blog Garden Rtc.blacker, support the original, reprint must explain the source, welcome to pay attention to the individual public number blacker

－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－

This period of time in Beijing for about ten days, relative to Shenzhen, this side a bit cold, also relatively dry, during the two events occurred:

1, Youku and potatoes were Ali pacified.

2, the Sohu producer Assembly held a

The first thing is a big deal, but it is not very important to us, the second thing is not a big deal, but it is very close to us, for the following reasons:

1, Youku Potatoes were acquired by Ali, the Internet is a bit cold in the winter, they want to find a big tree on the very normal, and we have nothing to do.

2, Sohu Producer Conference Many people do not know what is? And I don't know what's the matter with us? Let me explain it in a straightforward language from a commercial and technical point of view:

－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－

Youku potatoes were pacified, le Video and Sohu videos These small partners days is actually not good, because no profit, I analyzed their main cost of three dollars before .

1, Internet tariff

2, labor cost

3, video copyright

This part of the last in the WIN10 upgrade why to peer inside I also analyzed, and listed the pleasure of the financial results, see:

－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－

In the absence of more profitable channels (see ads We wearies, do members we too expensive), they can only choose to throttle.

But throttling to be targeted, the network tariff is certainly not down, the reason you understand, the human cost is also down (audio and video senior engineer to rob), that finally can only from the copyright.

There are currently two sources of copyright:

1, buy other people's (such as the United States drama, the cost is scary)

2, homemade play (such as the cock silk man and some users upload themselves)

American drama, domestic drama these royalties are hard to come down, its own cost is so high, plus a variety of approvals, all kinds of unspoken rules, say to a bitter tears.

There is only one way to solve this problem: "To improve the proportion of high-quality homemade dramas".

But here's a question: what happens when different users upload things that might not be the same level, and most of the time these things may not be high?

They think of the way to improve the user's production level, while encouraging users to provide more original things (to encourage the white is to give money), this is the purpose of the Sohu Producers Conference.

Speaking of which, you will ask: "Blacker, I still do not understand the producer of the Conference to make money with us what is the relationship?" ”

You think: If your self-created this thing to Sohu video up, click-through rate is very high, Sohu video will not provide you with some? Aren't you the equivalent of a producer?

You will also ask: "Blacker, the Chinese piracy so serious, I photographed those" collector's Edition "was pirated? ”

Ask very well, below I from the technical angle to help you analyze Sohu Video How to solve this problem?

Early technology: MD5 algorithm

Effect: as long as two video files are not the same, their MD5 value is not the same, if the same, the background is easy to find out.

Principle: Each video file generates a MD5 value, the MD 5 value of different files is impossible, once the same, the background database will be able to find out immediately.

Defect: want to pirate people holding your video file add subtitles, or a little clip, MD5 value changed, and then upload backstage on the detection, naked piracy.

－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－

Today's technology: Sohu video Gene algorithm

Effect: The background can automatically identify two similarity in 60% above the video files, once found pirated, immediately issued a warning.

Principle: Image recognition and big data combination, as follows:

1, each video file extracts the keyframe (not the key frame inside the video encoding, but the keyframe of the scene toggle)

2, compared with the video to compare the key frame comparison, is mainly compared to the waveform generated by the pixel, plainly, is the image recognition.

3, not much to say, some know not to say, do not know how to say (pro, pay attention to the public number Bai (blacker), you can also scan the QR code below).

4, if the similarity in 60% above the description is pirated (this ratio can be adjusted).

－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－

Analysis Here you know this producer of the meeting with our personal relationship!

As long as you can produce what other people want to see you can make money, because you take something to pass up and someone will give you money!

Of course, Sohu Video is only a typical case, if you can produce high-quality video content, who will provide you with a split.

Finally also remind those who have high-quality content but lack of mature technology friends, can consider I said above these things, focus on their business, first make money, video technology This thing, the threshold is relatively high, not 1:30 will be able to fix, this is one of the reasons blacker in the lake.

Like series of articles please pay attention to my public number blacker, or scan the QR code below (attention can send Sohu video one year membership coupon code (private messages to me), to force it! ）：

－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－－

This article from the first personal public number, see from the commercial and technical perspective of the video industry opportunities, support original, reprint must explain the source.

Android IOS WebRTC Audio and Video development Summary (48)--the opportunity to see the video industry from a commercial and technical perspective