Seedance 2.0 On LibTV it has fully launched.
So the Star Video 2.0 I used for 2 weeks turned out to be Seedance 2.0 itself!Hands-on with LibTV's new model Star Video 2.0, short drama creators' private Hollywood is here!
I have to say, Seedance 2.0's subject consistency, high-difficulty camera movement, adherence to physical laws, and voiceover are really on a different level from previous AI video tools, but on the official site a single video requires queuing for several hours, which is too hard to bear.
Based on hands-on experience,LibTV the overall video production experience is smoother, and the speed is much faster too, one video in a few minutes, no need to queue at all~
01. Hands-on with Seedance2.0
We open the LibTV official site, and in the video node, we can select and use the Seedance 2.0 video model.
Official site:https://www.liblib.tv
We only need a simple description to generate a complete video containing storyboards, transitions, and integrated audio and visuals.
An anthropomorphic kitten, at its own little stall, skillfully wrapping wontons, boiling wontons, scooping them out with a skimmer, packing them up and handing them toward the camera.
LibTV took less than 5 minutes to generate a 15-second video, with no need to queue at all~
Seedance 2.0 automatically fills in the visual details for me, adds coherent camera push-pull movement, and the sound of boiling wontons and the background music are generated in sync with the visuals,which is much simpler and more practical than before when first-last frame generation still required dubbing it yourself.
Next, I will use some common application scenarios for AI video to walk everyone through testing LibTV's Seedance 2.0玩法.
AI comics
This past half year has been the boom period for AI comics. Previously, making comic videos required doing your own storyboards, piecing them together segment by segment, lip-syncing...
With Seedance 2.0, we only need to provide detailed shot requirements, and it will break down the plot structure itself and generate coherent multi-shot visuals.
A high-energy xianxia battle shot, warm gold-red tones, 0-3 seconds: low-angle close-up of the protagonist's blue robe hem fluttering wildly in the heat wave, both hands gripping a giant sword with thunder patterns, the blade continuously flashing with crimson lightning,the ground lavabubbling and surging, in the distance demon soldiers roar as they charge closer, the protagonist says in a low voice, "Today, with this sword, I shall suppress your evil spirits!", accompanied by the sword's hum and the gurgling of lava; 4-8 seconds: a fast-cut orbiting pan, the protagonist spins and swings the sword, the blade tears the air and shoots out red shockwaves, the front row of demon soldiers is sent flying and shattered into ashes, accompanied by the whoosh of sword energy and the miserable howls of demon soldiers; 9-12 seconds: a low-angle shot pulling far away with freeze-frame slow motion, the protagonist leaps into the air, the blade gathers giant lightning arcs and slashes toward the demon soldier horde; 13-15 seconds: a slow push-in close-up of the protagonist landing and sheathing the sword, the robe hem rippling slightly in the aftermath, and says coldly, "The gate to this realm shall not be crossed," with the sound effects settling into lingering vibrations and fading wind.
Professional terms in the prompt such as low-angle close-up, fast-cut orbiting pan, and low-angle shot pulling far away with freeze-frame slow motion can all be understood by Seedance 2.0 and accurately presented.
AI simulated real-person short dramas
After Seedance 2.0 became explosively popular, because the generated videos were too realistic, content generation with real-person likenesses was immediately banned. When we use image-to-video in LibTV, we can first click Seedance 2.0 compliance verification, and only if it is determined not to be a real person can the video be generated.
We randomly generate a character image. Right-click the image, select Seedance 2.0 compliance verification, and after a few seconds, the verification result will appear.
LibTV also supports extrapolating a single image into four-panel and 25-panel coherent storyboard frames, helping us further plan and present the plot direction in a more complete way.
The generated storyboard characters and scenes have very strong consistency, and even if the movements are very large when making the video later, the visuals are not likely to break down.
The male lead walks forward alone, black-clad assassins behind him simultaneously draw their blades and close in on him, the male lead quickly reacts and repels the assassins.
Seedance 2.0 synchronously generates dialogue, ambient sound, action sound effects, and background music, so there is no need for later dubbing and editing, and the video feels very cohesive.
Short drama dialogue
For character images that have been verified on the canvas, right-click again to save them to My Assets, so they can be conveniently called up next time you make a video.
This way, the same AI actor can continue to appear in dozens of short drama clips.
Let's generate two clips and try it.
Clip one:
Cinematic realistic-style short drama, cool-toned business office lighting.
0-4 seconds: Full overhead shot, a luxurious conference room with a long table lined on both sides with executives in suits, the atmosphere is tense, the male lead @Image 1 sits in the head seat, twirling a pen with one hand, expression arrogant and contemptuous; opposite him, the female lead @Image 2 sits quietly, holding a kraft paper document bag in her hand;
5-9 seconds: Slow push-in medium shot, the male lead leans sideways against the chair back with his legs crossed, the corners of his mouth curl upward as he sneers, "Just you? You think you deserve to sit at this table? My company is valued at three billion, you can't even come up with a fraction of that," his tone is arrogant, his fingers tap the table making a da-da sound, the surrounding executives snicker;
10-15 seconds: The female lead unhurriedly opens the document bag and pulls out a stack of documents, gently pushing them to the center of the table, a close-up shot sweeps across the first page of the document, which clearly reads "Equity Acquisition Agreement" and her signature, the female lead raises her head and looks directly at the male lead, her gaze calm and sharp, the corners of her mouth slightly raised as she says softly, "This three billion, I have already paid".
Sound effects: the low hum of the air conditioner in the conference room, the sound of the male lead's fingers tapping the table, the rustling sound of documents sliding across the table, and when the last line lands, all sounds suddenly stop.
No text, watermarks, subtitles, or logos of any kind.
Transition point: The ending image of this segment: the female lead calmly looks directly at the male lead, holding the acquisition agreement in her hand, across the table the male lead's expression begins to crack from arrogance, and the conference room is dead silent.
Continue using these two character assets and extend the video.
Clip two:
Extend @Video 1 by 15 seconds. 0-5 seconds: Close-up of the male lead @Image 2's facial expression changing dramatically from contempt to shock to disbelief, pupils sharply contract, lips tremble slightly, he suddenly grabs the documents and flips through them, his hands are shaking, cold sweat seeps from his temples, the surrounding executives' smiles freeze as they look at each other in dismay; 6-10 seconds: Orbiting shot pushing in to a half-body close-up of the female lead @Image 3, the female lead slowly stands up, straightens her hair with one hand, fully commanding the room as she looks down at the male lead from above, her tone calm but utterly crushing, "From today on, every floor of this building is mine," while gently placing a company ID badge on the table and pushing it toward the male lead; 11-15 seconds: Low-angle upward shot, the female lead turns and strides toward the conference room door, the sound of her high heels is especially crisp in the silence, the male lead slumps in his chair with hollow eyes, the moment the female lead pushes open the door she turns back, side light falls on her face half bright and half dark, she says lightly, "Welcome to my company," and the door slowly closes behind her.
Sound effects: the hurried sound of pages flipping, the crisp echo of high heels on marble floor, the sound of pushing the door, and after the final line, the dull thud of the door closing. No text, watermarks, subtitles, or logos of any kind.
The plots of the two clips can be said to connect seamlessly, and the characters and scenes are fairly consistent, making them very suitable for AI short dramas.
AI Product Ad
White peach fruit drink product ad, splashing water, fruit and particles effects, fresh pink and white color palette. 0-3 seconds: close-up of the white peach flavored fruit drink bottle from @图片1 standing on a pure white surface, suddenly a ring of sparkling water splashes outward from the ground upward, water droplets refract rainbow light spots under the lighting, and a thin layer of mist condenses on the bottle surface; 4-8 seconds: orbiting shot, the camera rotates a full circle around the bottle, behind the bottle a massive torrent of white peach flesh particles bursts and pours down from above, pink-white flesh granules, translucent peach juice droplets, and crushed ice scatter in slow motion, some juice slowly flows down the bottle wall; 9-12 seconds: push in to a close-up of the bottle mouth, a plump white peach falls from above the frame into the bottle mouth, the moment it enters it stirs up a huge ring of water splashing outward, peach juice droplets and flesh fragments freeze and suspend in the air, and blurred peach blossom branches appear behind the bottle; 13-15 seconds: slowly pull back to a wide shot, the bottle stands steadily in the center of the frame, surrounding water, flowers and fruit particles slowly drift down like a cherry blossom rain, and from the side-rear of the bottle a soft warm light shines through the mist formingTyndall effect, the image freezes. Sound effects: the crisp sound of splashing water, the clinking of ice cubes, the whoosh of juice splashing, and at the end the dripping echo of a single water droplet falling. No text, watermarks, subtitles, or LOGOs allowed.
The fluid dynamics of the water splash, the texture of ice cubes and fruit collisions, and the Tyndall effect of light passing through the mist—the rendering quality fully reaches the usable standard of commercial advertising.
All-Around Reference
In LibTV's node workflow, Seedance 2.0 demonstrates its strongest all-around reference capability.We can connect various nodes together and feed it images, videos, audio, and text all as reference sources.
We can quickly make AI imitate the color style of any video, complex camera movements, or specific character actions, replicating the image in your mind with one click.
I used AI to replicate that stunning flying apsara neon dress dance from Where Winds Meet, the dance moves are super beautiful.
Refer to the camera movement of @视频2 and use @图片1 to replicate this video. Keep the scene and characters unchanged, a girl dancing on a mountain peak, with a highly Eastern artistic aesthetic
About Pricing and Perks
The scariest thing about making AI videos is burning money on gacha rolls. Running Seedance 2.0 on LibTV is not only fast, but currently also extremely cost-effective.
Buying a membership gives you free credits for Seedance2.0, Kling 3.0, and Kling O3, up to 240 clips~ the perks are worth thousands of yuan, very generous.
02. Agent / Skill Invocation, Automated Creation
If you don't even want to click on the canvas or connect lines yourself, you can also send your content creation requirements directly to Lobster. Libtv has packaged its content creation capabilities into Skills, so our Lobster can also automatically generate high-quality videos.
We open any "Lobster" and first have the Lobster install the LibTV API Skills via ClawHub with one click:
Install this skill:https://clawhub.ai/haofanwang/libtv-skill
Click the avatar in the top right corner of LibTV to create an Access key.
Send the Access key to the Lobster, and the Lobster can generate all kinds of videos.
Help me make an animated short film called "Little Snowman's Adventure," 15 seconds long, 16:9 widescreen.
The Lobster will call LibTV to create a project and generate the video, delivering the final cut directly to me, very convenient.
Calling the Skill to generate a video will use the Seedance 2.0 model by default. If you want to switch to models such as Kling O3 or Kling 3.0 for generation, you need to specify it via a command.
03. Some Sharing
The hard power of Seedance 2.0 is beyond doubt. We can see that video creation today is no longer simply gacha-style generation; it also includes AI's deep understanding of and reflection on creator intent.
Breakthroughs in underlying technology have given Seedance 2.0 extremely strong generalization capabilities and practical value. Whether it is a whimsical short video for personal entertainment or demanding industrial-grade commercial content creation, ultra-low-cost implementation is now possible.
LibTV, together with Seedance 2.0, provides short drama creators, advertising directors, and every self-media creator with a highly controllable, queue-free, film-and-television-savvy, efficient content workflow.
AI video is moving from a novelty toy to industrial-grade productivity.
Official website:https://www.liblib.tv
Original link:Seedance 2.0's queue-free, full-power version launches on LibTV! Produce an AI blockbuster in just a few minutes