'Midjourney in the field of video'! AI video generation rookie Gen-2 internal beta work leaked, netizens said it was too realistic

Gen-2, the AI ​​video generation tool that claims to be able to shoot blockbusters in one sentence, has begun to reveal its true colors.

A Twitter blogger has taken the lead in qualifying for internal testing.

This is the result he generated using the prompt word "a well-shaped or symmetrical man being interviewed in a bar":

I saw a man in a dark shirt looking straight at him While talking to the other person, his eyes and expressions reveal seriousness and frankness, and the person opposite him nods in agreement from time to time.

The overall video is very coherent and the picture quality is very clear. At first glance, it feels like a real interview!

Another version of the same prompt word generation is not bad:

This time the camera is closer, the background is more realistic, and the characters are still expressive.

After reading this set of works, some netizens exclaimed:

It’s incredible, just relying on text prompts can produce such results!

Some people said bluntly:

This is the arrival of Midjourney in the video field.

Blogger's actual test of Gen-2

This blogger is named Nick St. Pierre, who specializes in sharing his works made with AI on Twitter.

In addition to the realistic style seen at the beginning, he also released a set of Gen-2 science fiction works.

For example, "Astronauts Travel Through Space":

"An armed soldier runs down the corridor of a spaceship, a dark shadow destroys the wall behind him" :

"A family of microchip-eating robots in a human zoo":

"An army of humanoid robots colonize the frozen plains":

(It has the momentum of the White Walker army attacking the Great Wall in Game of Thrones...)

"The last man on earth watched the invading spaceship land over Tokyo. ”:

……

All of the above can be completed with just one prompt word, without the need to refer to other pictures and videos.

Although the effect of this set of science fiction is slightly inferior to that of "Bar Man Being Interviewed", the magical thing is that it can match the "chaotic" flavor of the AI ​​image generation model at the beginning. It's almost the same - it seems like I can see the shadow of AI such as Stable diffusion at that time.

For example, Nick St. Pierre said:

Gen-2 is still in its infancy, and it will definitely be better in the future.

We also found some results on the Internet tested by other people who have qualified for internal testing:

##It can be said that except for the more difficult science fiction works, it can handle everything else All pretty good guys.

AI video generation is also booming

The AI ​​tool Gen-2, which specializes in video generation, was released on March 20 (paper March 11).

As a newcomer in the field of AIGC, its iteration speed and quality are also quite fast:

The Gen-1 version was just born in February, and it can only be used for Already have videos for editing;

Now Gen-2 can directly generate videos using text and images as prompt words.

Officially calls it "the next step in generative AI", and the slogan is also very domineering:

say it, see it. (As long as you can tell it, I can let you see it)

The update of Gen-2 brings eight major functions in one go:

Venture video, text reference image generation Video, static image to video conversion, video style transfer, storyboard (Storyboard), Mask (such as turning a walking white dog into a Dalmatian), rendering and personalization (such as turning a head-shaking guy into a turtle in seconds) ).

Its appearance, like AI painting, allows people to see the huge changes hidden in the fields of film and television, games and marketing.

The development company behind it is also worth mentioning, it is Runway.

Runway was established in 2018. It has provided technical support for the special effects of "The Instant Universe" and also participated in the development of Stable Diffusion (a promising stock).

Tips: Gen-1 can already be played (after 125 opportunities are used up, you can only pay monthly), Gen-2 has not yet been officially released to the public. .

In addition to the Gen series, Microsoft Asia Research Institute has also recently released an AI that can generate super-long videos based on text: NUWA-XL.

With only 16 simple descriptions, it can get an 11-minute animation:

Move forward a little Son, on the same day that Gen-2 was released, Alibaba Damo Academy also open sourced the text-to-video AI with 1.7 billion parameters:

The effect is Aunt Jiang’s:

……

It can be predicted that not only image generation, but also the video field will become lively.

Ahem, so will it be the next wave of mass AI carnival?

Reference link:
[1]​​​https://www.php.cn/link/4d7e0d72898ae7ea3593eb5ebf20c744​​​
[2]​​​https://www.php.cn/link/e00944d55e6432ccf20f9fda2492b6fd​​​
[3]​​​https://www.php.cn/link/ce653013fadbb2ff27530d3de3790f1b​ ​​
[4]​​https://www.php.cn/link/6e3adb1ae0e02c934766182313b6775d​​​
[5]​​​https://www .php.cn/link/b9b72b29352f3764ea4dec130772bd9d​​​
[6]​​​https://www.php.cn/link/79d37fb2893b428f7ea4ed3b07a84096​​

The above is the detailed content of 'Midjourney in the field of video'! AI video generation rookie Gen-2 internal beta work leaked, netizens said it was too realistic. For more information, please follow other related articles on the PHP Chinese website!