Seedance 2.0 The creation of the world ' s wide-ranging videos is impressive, and it is no exaggeration to say that it will bring a major earthquake to the film industry and, of course, new opportunities to a wide range of users。
Now, everyone can bedirector (film etc).
But how do you do that
Cut the crap and just start。
Seedance, what is it
Seedance is an AI video generation model, 2.0, the most recent version of it, and the most powerful version of the entire AI video world。
But it's purely domestic, not foreign, and it's byte, the same dad who's shaking, so there's no need for magic on the Internet。
WE CAN BE USED IN THE OFFICIAL NETWORK OF DREAMS (AND BYTES OF PRODUCTS) OR IN THE APP
🔗 address: https://jimeng.jianying.com/
Figure: Use of Seedance 2.0 on Dreamnet
IN ADDITION TO THE WEBSITE, SOME OF THE USERS CAN SEE IT IN THE "DREAM AI" APP OR "THE BEAN BAG APP."。
THE OTHER BYTE PRODUCT, LITTLE SKYLARK, CAN ALSO BE USED, EITHER BY APP OR BY THE WEBSITE:
🔗: https://xyq.jianying.com/
💡 Seedance 2.0 is still in the grey scale test, whether it's dreamer net, or dream APP, or bean bag, and it's not all open, so there's no rush to see friends。
Seedance is not free, but users send 60 points a day for login. This 60 points can only generate a 10-s video。
Then the video will be generated by buying a dream member, and you can buy it according to your needs. (i.e. dreams are not just used to generate videos, but also pictures, digital people, etc.)
10 points = 1 yuan。
in other words, it takes $6 to make a 10-s video, of course it's the most basic case, and if you add, for example, a video reference (to be mentioned later), the score will soar to 130。
so, generating a 10-s video takes 13 bucks。
Seedance 2.0. What can you do
Seedance 2.0's ability to read the article I wrote two days ago:
image 2.0: The "ChatGPT Time" of the film is really coming
Seedance 2.0, how strong is it? I've sorted out these "shocking" cases and suggested a direct collection。
In short, it can make a court-wide video。
It is not an exaggeration to say that if it lets the restrictions go, it can wipe out the entire film industry。
Core game: full reference
How to use it is the focus of the article, but it's a professional threshold in itself, and it's too much of a Seedance 2.0, and it's hard to get it all out。
So I'm here to focus on its core function: full reference。
We used to generate videos in a text, or by reference to a picture, but now, Seedance 2.0 can refer to text, pictures, videos, even audio, and not just one。
It's very simple to use a reference, whether it's a reference or a video, you just have to upload them and then put them in a hint。
Figure: Use "Alternative Reference" in Seedance 2.0
In this case, for example, the complete phrase is:
@photo1 In the streets of Hong Kong, there was a fierce fight with men in black, who ended up falling down. Finally, @photo1 Shouting: "Who else?"
You can look at several pictures, videos, for example, six pictures, one video:
figure: reference in seedance 2.0
Specific reference can be made to those formats and how many references can be made to the description of the official document (link at the end of the text)
What use is reference
You might ask what reference to do or when to use it
Full Reference is one of the most powerful functions of Seedance 2.0, which can do too much。
For example, by looking at the image of a person, you create a video that makes the person in the video you want。
The “consistency” of the person and the scene is a basic requirement for the video to be made by reference to single or multiple maps。
Simply put, by reference to the video that you generate, you use the person and scene you've designated。
it's like you're the director, you pick an actor and let ta have to shoot in the studio you built。
but it is by no means limited to the "alternative reference" of sightance 2.0, which is one of the weapons that it can hang other video models。
It can also use an action to generate a kung fu film, a dance。
We can also use the mirrors to generate the same cool video。
You can use a spectroscope, a spectrograph to get a good video。
It's even using a music rhythm to make a card video。
You can see, "All-A-Reference" is actually a combination of fists: you can look at the image of the person, the scene, the action of the video, the mirror, the sound of the audio, the rhythm, etc。
This case, for example, refers to a character image, five costumes, a musical rhythm
Figure: Cases with reference to multiple elements
If the reference pictures and videos are not shocking enough or fresh enough, then the reference audio really surprises me。
It can be said that the "All-powerful reference" not only controls the resulting video through the image, scene, action, effects and sound of the person, but also directly "repeated" almost any video。
Once again, Seedance 2.0 is playing "doing the same" to the extreme, and this is going to be a double wave。
Voice and line: Finally
No sound, no good play。
Seedance 2.0, although first-time "single sound" on the line, has been effective enough to hit other products on the market。
The so-called simultaneous sound refers to the sound (including sound, white, etc.) generated with the video, rather than being added to the later stage alone。
FOR EXAMPLE, THE FOLLOWING EXAMPLE, WHICH INCLUDES WHITE, BGM, SOUND EFFECTS, ALL OF WHICH ARE DIRECTLY GENERATED, AND THE VARIOUS VOICES ARE VERY, VERY HARMONIOUS, RATHER THAN VERY ABRUPT, AND IN SOME CASES HAVE A WEAK EFFECT。
(The phrase "too close" is rather harsh, because that's what the original comics refer to.)
But this video was completed with the simplest hint:
Draws @photogram1 comics in a left-to-right top-to-down order, keeping the characters' lines consistent with the pictures, spectroscopy-to-spectrum and focused plottography into special sound, and the whole style is so sweet and sweet; by way of presentation @video1
in addition to basic linguism, seedance 2.0 supports many languages and even many dialects, including sichuan, chinese, etc。
Horriblely, it even supports drama. (preparation, Kyoto)
So you can say that Seedance 2.0 is a new height for the same time。
What's the hint
Although I have shown some hints before, especially how to use it as a full reference, this is not the whole point of writing。
Seedance 2.0 is smart enough to produce a good video, either through the simple words you used in front or through the complicated ones。
A simple reminder is, like we usually describe something, to make it clear, for example, that a girl is in an elegant tan and then takes another one out of a barrel and shakes her clothes. This is an example from the official case
If you want to be specific, you can spell out the content, the sound and the white of each image
Three seconds: the man in the suit sits in a bar, looks calm, shakes the glass. The camera is moving slowly, the light is advanced, the atmosphere is serious. The sound was low, and the man in the suit whispered, "This is a big deal."
3-6 seconds of footage: The woman behind her looks nervous and asks, "How big?" The man in the suit looked up and down the sound, "very big." Scratch hand close -- he put the glass down and the air full。
6-9 seconds of footage: suddenly the man in the suit pulled out of the table -- a big bag of exaggerated snacks, and the sound of boom was all over the table。
9-12 seconds: The woman behind her had her hands between her waist, muscles from stiff to loose and the whole person's face relaxed. The scenery is relaxed。
13-15 seconds: The man in the suit takes out a pack of snacks for women, pulls the camera out to show the whole view of the tavern, and the picture becomes blurry -- the subtitle pops up, "Be busy, have some snacks."
And if you want to be more professional and more manageable, how do you write the hint
In fact, there is an all-powerful formula:
Subject + scene + action + lens language + style
But the point is to be clear: who does what (scenario) does what。
A little strange about camera language? Of course, it is a little professional to be clear here and may include the entire audio-visual language。
But we don't know so much. We just need to know:
I don't need much explanation
Scenes: What's used to be called "thrust-and-screech" is a step-by-step slide, a step-by-step pull shot, a go-to lens or a first view lens, or a famous Hitchcock mirror, and is it slow? Or hand-held shooting
The prompt words are as follows:
A professional car-racing movie preview。
0-3 seconds: close-up shot. Photo @2, Racer's eyes are sharp, the dashboard pointer hits the red zone in a flash, and the hand is changing fast. Sound: Engine boom。
4-8 seconds: High-speed and filming. Photo @ 1, Yellow Runners drifted at extreme speed at the turn of the track, with a large volume of white smoke rolled up and the back of the car. A black racing car was close to trying to cross。
9-13 seconds: Slow motion (upgrade). The car went up the ramp and flew off. The camera rotates around the stationary body at 360 degrees in the air, showing the hanging details. The setting is sunset。
14-15 SECONDS: NORMAL SPEED. THE RACERS ARE DOWN AND RUNNING THROUGH THE FINISH LINE. BLACKSCREEN。
Is it suddenly a bit of a professional。
In addition to the lens, style is a key factor influencing the video. Like Miyazaki's Animation style, Tianxing Sun, Sabourpunk or Wang Jia-Hang
Is it a classic orange or a obscurant yellow
This example comes from the network, and its style is:
Hong Kong Art Cinema, 90s Hong Kong Film, High ISO Grain, obscurantious yellow and green (Yellow-Green Tint), frameding, depression。
With this, you can make enough videos to be professional, and of course you can use more skills if you learn a professional audio-visual language。
AS FOR THE "STRIPPING QUALITY," A LOT OF PEOPLE WRITE "HIGH, 4K, SUPER CLEAR," AND I DON'T FEEL NECESSARY. IT'S VERY EARLY, AND THE SYSTEM NOW EITHER SIMPLY DETERMINES WHAT CLARITY OR CHOICE TO MAKE。
Seedance 2.0, for example, default is 720p, and even if you have specified 4K, it is impossible to produce 4K video。
WHICHEVER PART OF IT, THE POINT OF WRITING A HINT IS TO BE CLEAR AND SPECIFIC, RATHER THAN TO WRITE SOMETHING THAT IS VAGUE AND GENERAL, SOMETHING THAT ONE CANNOT UNDERSTAND, AI
And that's what it feels like to be in a hospital movie。
Final Thoughts
It's a well-established fact that Feedance 2.0 is already the world's greatest AI video model, and that it's a tsunami-like impact on the whole area of video。
It is a crisis and an opportunity。
In just a few days, I saw a lot of good work by my netizens, and I saw the light of the country, and I saw an endless explosion of civic creativity。
The time has come for everyone to make movies and short operas