China's AI video TOP3 is rolling out! 1 second less than 1 penny, 3D full of full texture

  It's the first time in the history of the world.

A few recent AI short films have exploded the web.

It's the same thing that happens when you're a teenager and you're a teenager and you're a teenager.

Tiger freaks, tortoises burning people, and children's fantasies. A familiar rough picture takes you back to childhood.

The kids got up, the footage fell on the old man next to them, and the mood slowly changed.

And this kind of expression is appearing in more and more paintings.

From 2D hot-blooded animation to 3D-style animation to clay with hand-crafted texture, play can burn, play can collect, and the lens has begun to take on a life of its own.

After the National Day, the newly released Vidu Q4 Preview by Digital Technology appeared in several different styles of short films: The Nightmare Monkey has a high-speed intersection with a red blade and a giant snake.

Scroll to the left and look.

Technological exploration is also extending to more rigorous production lines.

The upcoming theatrical release of the upcoming AI-native film Satellite: The Future Is Enough is supported primarily by Digital Technology.

Bringing new technology into the cinema is the way the film industry is moving forward. From Cameron's exploration of 3D and performance capturing, to Lee's attempts to shoot 4K, 120 frames per second, new visual and audio experiences are often achieved in one go.

These frontier explorations were not necessarily perfect or even controversial at the time. But it was a technological breakthrough that pushed forward the language and immersion experience of cinema. From the small screen to the big screen, the intersection of the middle screen, was an intermittent test of technological ability.

The upgraded Q4 Preview features character performances, cameras, and complex visual effects, particularly in 2D and 3D animated scenes, with an initial discount of 0.09 yuan per second.

So a very tempting judgment emerged: AI video is both good-looking and cheap.

A lot of ideas have stopped before the bills of repeated attempts to shoot. Generate a few times dissatisfaction, look at the surplus, count.

And Q4 Preview's work has already gained its own recognition. Once on the scene, it once again surprised the whole network.

There's action, there's footage, there's a god in the characters.

The "Vidu Smells" is spreading from the hot-blooded plays to Eastern fantasy, animation and subtlety.

Today's AI video tournaments are becoming more and more like Chinese table tennis.

The higher the number of players, the thinner the competition. The team's struggles raised the industry's overall ceiling.

Seedance 2.5 continues to delve deeper into the professional production process. It emphasizes the ability to tell 30-second narratives, accurate references and editing, add white module controls, and green screen editing.

The continuous coding of authenticity and precise controls around long narratives, key puzzles and polymorphic references allows the creators to arrange a play in greater detail.

Vidu Q4 Preview has put "transcendental performance" in the foreground: the characters can act, the footage can move, the special effects can burn, and with lower test shoot prices, put these capabilities into more people's budgets.

Three routes, one direction:AIThe video is moving from "generating images" to "directing content".

After all, China's entire content industry has changed dramatically.

Look at the country's manga. In 2025, the animated film 'Hang Kong 2 Madness' surpassed $15.9 billion in worldwide box office, topping the global animated film box office.

This victory proves that Chinese audiences are willing to pay for top animation, great characters and great stories.

But what the audience remembers is that it's never just the special effects.

From the fairy-tale fairy-tale beach to the fairy-tale fairy-tale fairy-tale, the country is full of memories of both the big scene and the characters' temperament, emotions and personalities.

Technology is responsible for creating phenomena. Stories are responsible for keeping an audience.

At the same time, short and comic strips are exploding: close to 1.1 billion viewers across the country; the commercialisation of comic strips is accelerating, and industry policies are beginning to take hold.

It's getting bigger, and the audience is getting more provocative.

Shorts and comedies are saying goodbye to rough production and moving towards sophisticated competition.

It's just a ticket, it's a picture, it's a ticket, it's a play, it's a shot, it's a story, it's real.The role of AI video models in driving quality content creation is reflected in these specific footage.

The characterization, narrative synthesis and overall content performance of a work began to directly affect how far a work can go.

AIThe first half of the video is technical, and the second half is creative.

The opening of the exhibition is filtered, the camera is drawn closer, and the Q4 Preview is upgraded to a play on words, a jump, a big trick.

The characters' eyes, movements and voices are more specifically expressed.

The woman on the train holds a blue envelope, from tight to uplifting her eyes, to a grinning smile, and the mood unfolds slightly with the small stop outside the window.

There's no denying that Q4 Preview brings out the characters' hearts through their eyes, nose, and hands. Changes in faces, movements in their hands, and emotions follow.

"The characters have a God" and it is necessary to let the expression and the body accommodate the emotions.

Q4 Preview reinforces the characters' performances, and in such footage, you can see the proportion of fine performances.

The camera is powerful: jump over the roof, and you can also see the jump.

Not only that, Q4 Preview reinforces the interaction between dynamic lenses, cutting lenses and multi-scene villas. How the characters move, where the lenses look, can be a creative arrangement.

In this section of the roof post, Q4 takes a look at the level of the movements of the runner, the runner, the jerk, the knee, and the runner.

The full-body lens on the side alternates between jump points, roof gaps and landing points so that the viewer can see how the person moves from side to side.

In the anime special effects test, Q4 Preview puts the energy of the collision and the force of the character into the same shock: the starship hits the stones, the fragments are scattered on both sides, the gauntlet rises, the character retreats.

Then the spark spread, and the smoke rose, and the stars fled, and the figures and the golden light on the stone table dimmed.

The rise, collapse, and subsequent waves become a play, and complex visual effects can also play a part in the narrative.

It is worth mentioning that Q4 Preview supports a maximum of 15 reference charts, 3 reference audio, characters, scenes, props and character sounds, all of which can be used as creative input.

Our animated two-person dialogue test, using five reference charts and two tones, generates a 16-second Japanese conversation.

Two girls were sitting on the same bench, talking in their own vocal cords; one answered with their mouths open, the other with their eyes and faces.

Multi-figure references and audio references link the characters' performances, two anime characters, each with their own faces and each with their own voices.

Multi-figures give character shapes, tones give character characters, and performances bring out the relationship between the two.

These references allow creators to organize the characters and voices more clearly, concentrating the cue words to describe the lyrics, stops and emotions.

After viewing the episodes, you'll find a few very specific details: the characters wave their knives, the clothes roll with the shaking; the giant stones shatter, the fragments and the smoke then scatter; the energy dissipates, the characters and the light on the ground also weaken.

In order for a play to be credible, the model must combine the action and subsequent changes. The scenery changes its angle, and the characters, props and spatial relationships also keep pace.

This has led to a long-term exploration of biotechnology as a technological route.It's called the "World Model".

Along this path, Life Numbers has conducted world action model studies such as Motos, Motubrain, in particular the latter ranked first in the RoboTwin 2.0 and WorldArena rankings.

They integrate video, language, and action into a unified framework to link future image predictions to robotic motion generation.

Vidu is a product of content creators in a generic world model layout of living numbers.

If you compare a family of products to several children, Vidu is the "Doll" who has long stood before the creators.

The person's turn will not change his face, the props can not be connected through the shadows, and whether the next play can continue a state must be examined in the specific work.

The value of the foundation of the study is ultimately to turn it into a good lens in the hands of the creator. The character is more stable, moves more smoothly, control more reliably, so that "Vidu Smells" can move from a great sample to a continuous creative workflow.

And then there's another, more realistic book: How many more shots can you take with a camera like this?

The Vidu Q4 Preview has a price tag that is lethal in addition to the characters, the language of the footage and the visual effects.

Through API calls, 720P is available for about 0.6 yuan/second; 1080P is available for about 0.75 yuan/second; it is created directly on the Vidu platform, while the Preview phase offers a two-month limited offer.Initial discounts are down to 0.09 yuan/second.

From trial shooting to high-definition delivery, creators can choose the quality of the images as needed, leaving the budget for more attempts.

Imagine a 16-second battle of the martial arts.

In the first edition, the action is beautiful, but the force is not enough. In the second edition, the position changes, highlighting the moment. In the third edition, the character character is added, capturing the change of eyes.

……

In the past, every attempt could have meant extra manpower, time and production budgets. Today, at the minimum discounted price, the cost of five generations is only 7.2 yuan.

When it cost a few bucks, the creators finally had the guts to grind a lens repeatedly.

In addition to price, efficiency is equally important.

Q4 Preview is 16 seconds long, supports 2K, 4K output, personality and audio references are repeatable.

This means that creators can replicate characters and voice references, try new movements, expressions and positions in different shots, and reduce the burden of repetition.

Official data shows that a 5-second video can be generated in about 110 seconds at the earliest. The actual time is still affected by the resolution, generating specifications and alignment.

In the past, ideas had to wait for production. Now, production has started to catch up with ideas.

Multi-reference capabilities help creators maintain consistency of character and style. Quickly generating shortened feedback cycles. Low prices make repeat screenshots more affordable.

The three are superimposed on each other, directly changing the way the content of the product is produced.

There's a deeper technical logic behind this race: optimizing algorithms to reduce the computational costs of producing videos.

Prior to this, Biometric Technology and the University of Tsinghua jointly launched the TurboDiffusion study, which explored video generation acceleration from low-bit computation, scarcity of attention, and reduction of sampling steps.

It reveals a common direction for the industry:With higher computational efficiency, supporting lower production costs.

This is reminiscent of the impact DeepSeek has had on the big model industry.

As high-performance models become cheaper, technology has the opportunity to shift from the expensive tools of a few to the everyday productivity of more people.

AI video is also approaching this corner.

The most anticipated "DeepSeek Moment" is the cutting edge video generation capability for ordinary creators on a budget.

Back to cooking again! The kids are up, the kids are up, the mirrors are on the old man's face, and the story has a flavor.

For a few seconds like this, you have to polish your face, arrange a pause, and find the right time to cut the mirror. The model increases the performance of the character, squeezes the price of the test shot, and saves the budget.

One eye is out of place, you can try a few more editions; a play is not in the mood, and there is a budget remake.

The second half of China's AI video is here to bridge the gap.

Enter the exclusive invitation code: XINQ4, log in to Vidu.cn and get 500 points by registering a new user.