Domestic AI videos have become popular around the world, and foreigners collectively stand up! Fast to shocking, the shocking price is as low as 4 points

Article source: Xin Zhiyuan

Image source: Generated by AI

At the beginning of the new year, Vidu, who has always been low-key, released a big move.

Vidu 2.0, which was shocked and launched, was faster, and the video generation speed broke the record and ran into less than 10 seconds.

Not only that, it also reduces the cost-it only costs as little as 4 cents per second, directly entering the era of “minutes”.

Moreover, the effect of the model is also higher, with higher quality and a more stable style. Isn’t this the legendary saying of “increasing the quantity without increasing the price”? Vidu’s sincere users have actually received it.

Now, we only need to wait about ten seconds for Ultraman and Musk to make up!

It can be seen that in terms of character performance, Ultraman and Musk are almost consistent with the picture, and they smiled at the same time when shaking hands, following the prompts well. In addition, the background of the OpenAI office has also been completely restored.

Speed rolls across the sky

‍Since the birth of Sora, large video models have been iterated endlessly, but they have never been truly popularized.

One of the key reasons is that the “generation speed” is too slow-not only does the queuing time take a few hours, but the actual reasoning process takes a long time.

However, generation speed has always been Vidu’s specialty.

When Vidu 1.0 was launched at the end of July last year, the measured reasoning speed of a single segment fell within 30 seconds, making it the fastest in the world.

This time, Vidu 2.0 ‘s generation speed has broken through to seconds. Both graphic videos and first and last frame-generated videos generated 4 seconds of 512p single-segment video in less than 10 seconds, which is three times faster than the previous version 1.5.

True·Drinking water produces it!

If it is compared with other videos, Vidu 2.0 will directly crush it.

Netizens who were the first to get the internal beta commented on social platforms that Vidu 2.0 is better in terms of generation speed and quality.

It is not difficult to imagine that content production links such as advertising manufacturing, film and television, animation, and games will be completely subverted.

The traditional production process of advertising commercial films requires planning, shooting, post-editing, dubbing and other links. The shooting cycle is 3-5 days, and post-production is 5-15 days. A 1-minute commercial often takes 1 month to deliver.

The debut of Vidu 2.0 means that it only takes 5 minutes to run 1 minute of video material. It not only saves the real shot process, but also saves 90% of the post-process links, which is a perfect productivity winner.

Costs enter the era of “division”

While major AI video platforms are still taking advantage of users, Vidu 2.0 has directly pushed video costs into the era of “4 cents”.

You are right, the cost of a single second video is as low as 4 cents!

For users, adopting AI video tools hopes to reduce costs, but in the actual use process, they find that they are greatly discounted from expectations.

Currently, video models on the market have a monthly point limit.

Compared with platforms such as Runway and Sora, which often charge yuan per second, the price/performance ratio of Vidu 2.0 can be said to be crushing.

Specifically, the unit cost per second of Vidu 2.0 is as low as 0.04475 yuan/s, which is at least 50% lower than the previous generation version 1.5, with a single video architecture. Compared with its peers, the cost of Vidu 2.0 is nearly 40% cheaper.

Based on the unit price of the lowest version of each domestic product, Vidu 2.0 720p is only 43% of the industry average charge level. The industry average price is 0.6 yuan/s, and the Vidu average price is 0.2587 yuan/s.

For example, if we all adopt a 720p primary package, Sora can only generate 14 seconds of video material for 10 yuan, while Vidu 2.0 can generate 39 seconds of video material, which is more cost-effective than other homes.

In addition, Vidu 2.0 has also quietly launched the “peak shifting mode”.

That is to say, during the peak period of platform traffic, unlimited video material can be generated without consuming points at all. There is nothing more fragrant than the official letting you collect wool.

Film-level quality, the effect is as stable as an old dog

In addition to its fast speed and low cost, Vidu 2.0 has broken records in terms of generation effectiveness and stability.

It can even be said that Vidu 2.0 broke the impossible triangle of the video model industry in one fell swoop and achieved “fast, good and cost-effective.”

Without saying much, we will directly start a wave of actual measurement.

Picture generated videos, with high consistency in any style

The most concerned issue for AI video users is whether the style can always be consistent during the video generation process.

This is the core of the implementation of industrialization and commercialization content.

Next, I will test all contestants with the same question: “A curly man among flowers, Impressionist oil painting.”

After comparison, we can see that Vidu 2.0 is significantly ahead in style retention-the flowers in the man’s hands are always consistent with the flowers in the background of the picture.

Moreover, our personal test found that in the graphic video function, the style of real characters is also maintained very well.

Refer to student videos to unlock multiple tricks

As the world’s first AI video model to launch the “Reference Live Video” function, Vidu has always been in a leading position in the controllability of video generation.

This time, the upgrade of Vidu 2.0 will bring this advantage to the extreme.

Overall, Vidu 2.0 shows excellent stability when dealing with different styles of subjects and backgrounds.

Our actual measurement found that the effect of Vidu 2.0 is indeed amazing.

Upload photos of a girl, the Parthenon and a bouquet of flowers, and Vidu 2.0 seamlessly presents them naturally in one scene, with a natural and beautiful picture.

Then upload an orange cat, a yellow chicken, and a picture of the outdoor lawn, and tell the model to generate a video of two cute little cuties staying on the green lawn.

Vidu 2.0 really makes two little animals in the same scene, with a kitten chasing a chicken, come alive. There are also leaves swaying in the air with the wind, giving people an immersive sense of reality.

The transition between the first and last frames is silky

In addition, the transition between the first frame of Vidu 2.0 is also smoother.

This time, Vidu 2.0 seems to have come with some kind of lens “black magic”, making the switching between the first and last frames more imaginative and more storytelling.

Let’s try the difficulty by asking two ancient people to drink tea around the stove in autumn, and then instantly travel to the streets of the snowy ancient city.

Sure enough, this question stumped many video models.

There are problems that the switching is very stiff and the effect is stuck. Sora’s switching even directly into “PPT mode.”

In contrast, Vidu’s transition silk is poetic and ancient, and the entire lens is very stable when switching

To measure it, the person standing in front of the space-time tunnel suddenly traveled outside the universe.

Vidu 2.0 seems to be really controlling the time disk, making people lose themselves in the endless whirlpool of time and space.

Let’s take a look at a case that better illustrates the seamless connection between the end and end frames.

A picture of Iron Man and Spider-Man hugging affectionately is transferred to the same scene on TV. The color changes from dark to bright, which follows the style of the original picture well, reminding people of the “push” lens commonly used in movies.

Movement of camera, good performance

The common up-down, left-right, push-pull and rotating lens movements in movies are difficult to understand in ordinary AI video generation models, let alone smooth lenses.

In terms of lens movement, Vidu 2.0 not only has a better understanding of the user’s lens movement instructions, but also presents it more naturally and smoothly, like a master lens language.

Take a practical measurement, let the camera slowly advance towards a girl in a cafe, and then rotate it counterclockwise. Vidu 2.0 is done very well.

Following the prompt of “lens rotation counterclockwise”, the video of the ancient lady turned as required, making it very silky.

Foreign netizens received rave reviews

I have to say that Vidu 2.0 is really good at playing. Overseas netizens applauded the table after testing it.

The actual measurement of the first and last frame transitions, the dreamy effect is realistic, which is simply an amazing batch

Netizen “Lars_Pragmata” said that Vidu 2.0 ‘s compliance with prompts is amazing, the subject consistency function is crazy, and the generation speed is amazing!

Below is her own Code-Eter, which will be available on YouTube after being paired with background sound effects. With Vidu, it feels like everyone is a director. Whether it is an action movie, a science fiction movie or an anime movie, we can handle everything.

Netizen “Amir Kerr” wanted to generate a video of a rabbit holding a white ball with its head lowered, but no other AI could do it because they needed a face like a human.

But he was pleasantly surprised to find that Vidu 2.0 successfully did it.

Another netizen,”Naegiko”, said that Vidu 2.0 ‘s effect is comparable to Veo 2, and the Japanese comic short films produced are completely as good as the masterpieces.

Netizen “Taher Dhanerawala” exclaimed: Now we can make blockbuster movies with a single image. This function of transforming static visual effects into dynamic storytelling scenes is simply a unique one.

He has created three scenarios of the universe: the mysterious portal in the desert, the wanderer entering the surreal cosmic void, and the rebirth of the celestial palace floating among the stars. Later, Vidu 2.0 combined these three pictures into a science fiction blockbuster, with amazing results!

Breaking one million users in 20 days, Vidu creates a growth myth

On this highly competitive AI video track, Vidu is doing miracles.

With the release of Vidu 2.0, the company also announced its operating data for the first time, and its astonishing growth rate has attracted the attention of the industry.

Since its global launch at the end of July this year, Vidu has only taken 20 days, and the number of users has exceeded one million; in 100 days, the number of users has exceeded ten million.

So far, the platform has generated more than 100 million videos. Among them, the new function “Reference Student Video” has only been launched globally for three months, and the number of videos generated by users has reached tens of millions.

Compared with competing products, Vidu’s growth rate is far ahead in the world.

Benchmarking data shows that Aishi Technology announced on November 18 that the number of users exceeded 12 million, and Kelin announced on December 10 that the number of users exceeded 6 million.

As early as the end of October last year, Vidu had achieved tens of millions of users, showing strong development momentum.

At present, Vidu’s service scope has covered more than 200+ countries and regions around the world, and its user scale ranks first in the world. Users from different countries, ages and occupations use Vidu.

In Oklahoma, 60-year-old children’s writer MamaFish hopes to turn her fairy tales into vivid and interesting videos.

However, the high cost of professional video production has always been a threshold in front of her. Until I met Vidu, this dream finally found a breakthrough.

With the help of Vidu, this grandma who never understood video production was able to create 50+ exciting scenes in 20 minutes.

What is even more surprising is that her work “If I Were A Cow” has received a lot of favorable reviews on YouTube. It not only realizes her creative dream, but also brings considerable benefits.

As a veteran player in secondary culture, well-known creator@8co28 has deeply experienced many AI video creation platforms such as Gen-3 and Luma.

In the end, he chose Vidu as his creative tool.

The works created through Vidu have helped him gain more than 350,000 loyal fans on social platforms.

Vidu’s product power not only won the favor of individual creators, but also won recognition from enterprise-level customers.

Taking Evoke, an AI graphics video application, as an example, after connecting to the Vidu API, the model accuracy rate is increased by more than 60%, the video generation time is shortened from 270 seconds to 35 seconds, and the probability of users canceling video generation is reduced by 10%.

These significant performance improvements directly translate into a better user experience, helping Evoke achieve double breakthroughs in user scale and revenue in overseas markets.

Vidu’s success is not only a product victory, but also a demonstration of China’s hard power in technological innovation, so it has won the favor of users around the world

The launch of Vidu 2.0 is also a new milestone in the entire video generation industry. The “second-level era” of AI video promoted by Vidu 2.0 will also allow videos to be generated in real time. You can boldly imagine that new experiences such as interactive short dramas and interactive games may be born soon.