Shulou(Shulou.com)11/24 Report--

With the advent of the AIGC era, the whole field of content creation has been subverted! In the face of new opportunities, who will be replaced and who can overtake around the corner?

Recently, AI is running too fast!

Foreign bigwigs quarreled over a piece of AI non-Proliferation Treaty. On the domestic side, bilibili, iqiyi and other Internet enterprises also talked about AIGC at the Chengdu audio-visual conference. Among them, Chen Rui also introduced bilibili's AIGC content at the scene.

Editor be like 👆 Chen Rui, who stays up late to chase hot spots, said that with the advent of the AIGC era, each of us is in a new era of science and technology accelerating change:

Young people's demand for AI technology-related content is increasing rapidly, and more and more UP owners are also producing all kinds of ChatGPt and AIGC-related content. So far, the number of ChatGPT-related content played on bilibili has reached 200 million.

Chen Rui predicted that the advent of the AIGC era is not only an innovation in the field of science and technology, but also has a subversive impact on the entire Internet industry and even on the whole society.

In the era of explosive growth of AI technology such as ChatGPT, Chen Rui's judgment is not groundless.

Some time ago, the AI painting artifact Midjourney, which came out of the circle with "a pair of Chinese couples sitting on the roof", is not too hot.

After an influx of new users, the server was overwhelmed, forcing officials to stop free trials.

At the same time, all kinds of amazing AI paintings emerge one after another--

There are Musk who traveled to the Soviet factory, the younger sister who watched Haitang, the Chinese version of Hepburn, and so on.

Source: Cydiar share a prompt here, you can try:

A Chinese 20-year-old Woman, looking like Audrey Hepburn, Black hair, standing on 2023 Tokyo street, hyper realistic portrait photography, pale skin, dress, wide shot, natural lighting, kodak portra 800, 105 mm f1. 8 ar 32k-- ar 16:9-- v 5-- s 750-- Q 2

Moreover, such a shocking V5 has just been launched, and the more powerful V6, V7 and V8 are said to be on their way.

At the AI subversive Video creation Conference in Chengdu, Chen Rui mentioned that the development of AIGC will greatly improve the efficiency of content creators. Just like the advent of cameras on mobile phones in 2000, which allowed ordinary people to shoot at will, the emergence of AIGC also enabled many ordinary people to create content that only professionals could create in the past.

Some time ago, a foreign guy tried to create a video led entirely by AI. This video from script to director, screenwriter, photography, soundtrack work, AI all round.

And this video is made entirely of today's mainstream AI tools, and this little brother is actually just an AI tool person (by the way). In other words, we can now, like him, completely manipulate AI to create videos.

Among them, the script is written by ChatGPT, the image is generated by Midjourney, then Runway Gen-1 is used to apply the image style to the source video, and finally, the music is played by Boomy AI.

First, say to ChatGPT, "write me a script to generate a 9-second video consisting of three 3-second clips." The story is that a man in his living room wants to take science fiction as the theme. "

ChatGPT generates the following script.

The second step is to create a reference image in Midjourney according to the script.

Prompt is "imagine sci-fi film still, medium shot, centered, side-angle view, a man sitting on a chair, holding a glowing orb in his hands, living room, new york, 4k-- ar 16:9" (imagine a science fiction movie still, middle, center, side angle view, a man sitting in a chair with a glowing sphere in hand, living room, New York, 4k-- ar 16:9).

In the third step, the little brother rearranged some furniture in the living room to better match the reference image to get a clearer output.

Then he shot the original scene with an iPhone.

In the fourth step, he uploaded the video and reference image to Gen-1, which immediately turned the video into the style of the reference image.

In this step, I didn't use a text prompt, but added-- interpolate as a parameter for smoother output.

The fifth step is to import the video into the video editor.

He uses iMovie. After importing the video, trim some clips and connect them together.

In the sixth step, the little brother generated the soundtrack from Boomy according to the style and mood he wanted.

He chose a song called "Trippy Harvester by Peaky Chicky", which matched the atmosphere of the scene very well.

The last step is to add music and export.

Fortunately, not too much editing is needed, because the video matches the music perfectly.

In this way, after 3 hours, a dream blockbuster was born!

AI is in charge of manual labor, and I am in charge of brainstorming. Domestic bilibili UP owner "Mega can play", but also through several AI tools, let Jobs "come back to life".

After visiting Jobs' former home, Mega suddenly had an idea to generate a video of himself interviewing Jobs.

Jobs' brain, of course, comes from ChatGPT. Because it has been trained with 175 billion parameters, it can talk in the style of the corresponding character.

Jobs' voice, on the other hand, was generated using a Tacotron 2 model.

Mega processed and tagged Jobs' speech materials in segments, trained for a few days, and finally generated Jobs'AI speech model.

And then there comes the most interesting step-- cloning a person with AI.

If Master Qiao is still alive in 2023, he should be an old man in his seventies.

Mega uses Midjourney, adjusts various parameters and prompt, and finally produces an ideal final version.

In the following video, we see the "resurrected" Jobs describing his rebirth as follows:

It's kind of amazing to come back to the world in this way. I'm surprised to be reborn in AI, but it's just one of many incredible human advances.

At the end of the interview, Mega asked the question he most wanted to ask: how do you feel about Apple today?

Jobs' answer is full of bright spots: "I can't evaluate Apple's current culture and practice because I'm no longer a part of him." But I really hope that Apple will always remember the importance of innovation and breaking through the status quo. "

In addition to being able to generate what we want out of thin air, AI also has an important "mission": to get rid of the tedious "dirty work" of the production process.

About the famous UP master half Buddha fairy, there has always been such an unsolved mystery: why are there so many funny memes in his videos?

People are so curious about this question that this Zhihu question has already received 430000 page views.

Under this question, the half-Buddha fairy himself also appeared and answered the questions of the melon people-in fact, these materials were crawled out with Python.

Another half-Buddha fairy said that of course he had a team, otherwise according to this output, physics would have been subverted without a team.

Now, with AI tools, these emojis don't even need crawlers and can be mass-produced by themselves.

Although human flesh production is still needed in the process of integrating materials, finding suitable sound effects and pictures, and adding post-dubbing, it is likely that AI will be able to package all these functions for us in the future.

For example, recently, after ChatGPT is connected to the Internet, you can access a variety of third-party plug-ins.

One of the "code interpreter" plug-ins allows us to edit videos directly with ChatGPT.

As long as you give an instruction, ChatGPT will process the uploaded video file, and after a moment, you can download it directly. That's right, it just blew up.

From this point of view, with the blessing of AI in the future, human creators may really just need to come up with a brain, and those who do not want to do it, or trouble, can be left to AI to be responsible.

Hold still. When overtaking at the corner sees this, some people may exclaim that since the mode of knowledge supply has been completely subverted, isn't it true that the jobs of knowledge creators have also been destroyed?

In fact, it hasn't come to this yet.

It is far more practical to let AI become a sharp weapon in your hand than to worry about being eliminated by AI.

You know, many people don't have the ability to put forward their needs, and they simply can't figure out what they need.

For example, if I give you an AI drawing tool, what can you draw with it?

Its upper limit depends entirely on the upper limit of your prompt ability.

The ability of AI has been put here for you, and how you can control it and perform miracles with it depends entirely on your command ability.

For example, now, if you were asked to generate a picture of the port, how would you describe Midjourney and Stable Diffusion?

A real prompt master can write "harbor, ship, sunset, beautiful light, golden moment …" Surrealist, focused, detailed, etc. The hint of "Movie quality, Masterpiece" allows AI to draw a perfect painting.

One of the most impressive examples of the editor recently is that in Bing's Image Creator, I wanted to try to draw the scene of Homer.

However, when I want to enter prompt, I suddenly run out of words.

The editor has an idea, so let GPT-4 teach me?

After putting the prompt given by GPT-4 into Image Creator, the resulting painting really shocked my family.

The creator of the Space Opera House, the winner of last year's art competition at the Colorado Fair, directly refused to share the tips he used on Midjourney.

It is said that it took him more than 80 hours to complete the painting in 900 iterations.

This is why the annual salary of foreign prompt engineers has reached 360000 US dollars.

As far as AI painting is concerned, a good prompt engineer must have a multidisciplinary background. Among experienced engineers and amateurs, there is a clear threshold.

Only by mastering a wealth of professional knowledge such as art history and graphic design, can we write "shooting with 35mm film" and "Persia." Isfahan's architecture, "the style of French painter Henri de Toulouse-Lautrec", is a professional prompt that produces the following paintings.

From this point of view, the differences between people will be infinitely magnified in front of AI.

Obviously, you need to be a person with sufficient breadth and depth of knowledge, and AI is as good as you are.

When information can be obtained extremely freely, the reprocessing of information becomes more important, and at this point, we all stand on the same starting line.

It is conceivable that in this case, the payment for knowledge will be a big business opportunity spawned by the mobile Internet.

In just the past few months, we have witnessed the iteration of the speed of light in various AIGC products, making it easier than ever to become a "knowledge intermediary".

With the development of AIGC technology, making a video may not be much more difficult than writing an article. Content creators only need to come up with the core ideas, and the rest can be automatically generated by AI, which will greatly reduce the difficulty of video production.

The number of individual creators may also see rapid growth.

In particular, it may be the creators of bilibili's knowledge class UP. After all, it saves time to cut the video and to find information. As long as there are core creative points, making high-quality videos is not a problem.

At that time, the creative potential of high-quality creators will be fully developed, and video will become the absolute mainstream of content.

For creators, this is not some kind of distant prophecy, but a near future within reach.

That's why bilibili's ChatGPT-related content has exploded in recent months, and many UP masters have directly begun to try to use AI to do video.

Maybe it won't be long before we can see more and more independent creators appear. They no longer have to choose a team to create this heavy industry model, but one or two main creators + AI can travel light, create with ten or 20 times higher efficiency than in the past, and express the content that they do not have the ability to make into a video.

It is conceivable that with the prosperity of AIGC and the development of knowledge video platform, a rising tide lifts all boats.

So, in the near future, who will be replaced by AI, and who will overtake around the corner?

This article comes from the official account of Wechat: Xin Zhiyuan (ID:AI_era)

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.