Video model · ByteDance ·
The newest version is Seedance 2.5Seedance 2.0 is a video model by ByteDance. In Endless it costs from 19 credits per video and is available on the Hobby plan and up.
Seedance 2.0 is ByteDance's all-purpose video model for 4 to 15 second clips at up to 4K, guided by image, video and audio references.
From 19 credits per video
Seedance 2.0 takes text, images, video and audio in one model and generates sound together with the picture, including dialogue, ambience and effects. You can attach images, clips and audio as references and name each one in the prompt, such as @Image1 or @Video1, to say what it should control, like a character's look or a camera move. Clips run from 4 to 15 seconds at up to 4K.
Teams use it for short ads and social clips where a product or presenter has to look the same in every shot, and for turning a single still into motion. A reference video can supply a camera movement to copy and an audio clip can set the timing for dialogue or music, so a storyboard frame and a voice track can become one finished clip.
Seedance 2.5, its successor in the Endless catalog, reaches 30 seconds in one pass, takes up to 50 references and can edit or extend an existing clip. Pick Seedance 2.0 for clips up to 15 seconds at up to 4K, and switch between the two in the same workspace, in chat or on the canvas.
Seedance 2.0 takes text, images, video and audio in one model and generates the picture and the sound in the same pass. fal's prompting guide asks for six things in a prompt: who is in the shot, what they do, where, how it is framed, how it is lit and what it sounds like. Words like "beautiful" or "epic cinematic camera" name no object, light or move, and a prompt built from them comes back with the most generic reading of the scene.
A structure that covers most clips:
[Subject, concrete: who, wearing what, holding what].
[Action, with its physical result]. [Setting and light].
[Camera: where it starts, how it moves, where it lands].
[Sound: dialogue in quotes, ambience, effects, music or "no music"].
Write motion as verbs with a consequence. "She drops into a low spin, the skirt flaring" gives the model a path to animate; "a graceful dancer" does not.
Named moves read cleanly: dolly, pan, tilt, crane, push-in, pull back, arc, rack focus, locked-off. Add lens and framing where it matters, such as "a long lens", "low front angle" or "chest height". For each shot, write three things: the framing it opens on, the move, and the framing it ends on.
Seedance 2.0 can cut between shots inside one generation. fal's guide says writing the cuts out with "cut to" is more reliable than letting the model invent a shot list, and that three quick cuts fit comfortably in one clip. Repeat the subject's key traits after each cut so the same person or product carries through.
Generate audio is on by default and covers dialogue, ambience, effects and music. ByteDance describes stereo output with music, effects and voices as separate layers aligned to the picture, including small foley sounds such as fabric rustling.
Attach up to 9 reference images, one reference video and up to 3 audio clips, then name each one in the prompt by its tag: @Image1, @Image2, @Video1, @Audio1. Start the prompt by saying what each tag stands for, then write the action. ByteDance's own example assigns four images four roles: a storyboard, a character, a scene and the props. ByteDance says references can control composition, camera language, motion rhythm, effects and sound.
In your workspace, open Video and pick Seedance 2.0 in the model picker.
Write a prompt with the subject, camera and motion, or start from an image you upload or generate.
Generate, then refine in chat or on a canvas. Every clip is saved to your workspace assets.
Grouped by use case. Each one opens in Endless with Seedance 2.0, ready to edit and generate.
@Image1 is the soda can, exact red label and silver top, identical in every shot. Extreme macro of the can on crushed ice as condensation runs down the label, then a hand cracks it open and fizz sprays upward in slow motion. Slow push-in, hard backlight. The hiss of the tab and ice crunching, no music.
A matte black espresso machine on a concrete counter at 7 a.m., soft window light from the left. Locked-off close-up as a double shot pours into a glass cup, crema rising, then the camera pulls back as a hand lifts the cup away. Machine hum, the drip and a spoon against glass.
@Image1 is the lipstick, exact gold tube and berry shade. On a seamless pink backdrop the tube spins and uncaps in macro, cut to a model applying it in close-up with a slow rack focus to her eyes, cut to the tube standing in a field of pink petals. Upbeat electronic beat with a click on each cut.
A white running shoe lands in a shallow puddle on a city street at night, in slow motion. Water bursts outward and neon reflections ripple across the surface. Low tracking shot alongside the shoe as it lifts off again. Splash, footsteps and distant traffic.
@Image1 is the skincare bottle. A woman in her twenties holds it up to a phone camera in her bathroom, natural morning light, a touch of handheld. She says, "I've used this every morning for a month." She dabs a drop on her cheek and smiles. Room tone, no music.
Two-shot at chest height inside a cramped food truck at night. The cook, a woman in her forties with a red bandana, slides a box across the counter and says, dry and a little proud, "Last one tonight, and it's on the house." Cut to a tight single on the customer, a tired young man in a delivery jacket, grinning: "You say that every night." Sizzling grill and traffic outside, no music.
A detective in a rumpled gray suit crouches in a rain-soaked alley, a flashlight cutting through the fog. Slow push-in as he picks up a dropped umbrella and says under his breath, "She was here ten minutes ago." Steady rain on metal, a distant siren, no score.
A glassblower in a dim workshop turns a glowing gob of glass on the end of a pipe, sparks drifting. Locked-off medium shot. Without looking up he says, quiet and worn out, "Thirty years, and it still surprises me." The roar of the furnace and the hiss of glass in water.
A grandmother and her granddaughter fold dumplings at a kitchen table under warm lamplight. The girl's dumpling falls apart and the grandmother laughs: "Again, slowly this time." Close two-shot, slow push-in. Bowls clattering, rain on the window, no subtitles.
Immersive first-person ASMR shot of hands on a wooden desk: they peel tape off a cardboard box, unfold tissue paper and lift out a ceramic cup. Close-up, soft daylight. Crisp tape tearing, paper rustling and the light knock of ceramic, no music, no voice.
Wide shot of a lighthouse on a cliff in a storm, waves exploding against the rocks. Cut to a medium shot of the keeper climbing the spiral stairs with a lantern. Cut to a close-up of his hand pulling the lever. Cut to a wide shot as the beam sweeps across the dark sea. Thunder and howling wind.
Low tracking shot behind a mountain biker racing down a forest trail. Cut to a close-up of the front wheel kicking up mud. Cut to a first-person view over the handlebars as the trail drops. Cut to a side shot as she lands a jump and skids to a stop. Tires on dirt, heavy breathing.
Close-up of a chef's hands chopping scallions on a wooden board. Cut to a medium shot as she slides them into a smoking wok. Cut to an overhead shot of noodles tossed in the air. Cut to the finished bowl placed on the pass under warm light. Kitchen clatter and sizzling.
A flamenco dancer in a red dress stands still in a dark courtyard lit by one lamp. Cut to her heels striking the stone floor in close-up. Cut to a low front angle as she drops into a fast spin, the skirt flaring. Cut to a wide shot as she stops dead and raises her chin. Guitar, palmas and heel strikes.
1920s jazz club, warm amber light. Wide shot of couples dancing the Charleston on a crowded floor. Cut to a close-up of two-tone shoes kicking in rhythm. Cut to a slow arc around the lead couple as they laugh. Brass band, clapping and the crowd cheering.
@Image1 is the mascot, a round orange robot with one blue eye. @Image2 is the city park. The mascot rolls along a path in the park, waves at two joggers, then uses a telescopic arm to free a kite from a tree for a child. Wide shot, then a slow arc around them. Light orchestral music and playful beeps.
@Image1 is the presenter, @Image2 is the studio set, @Video1 is the camera movement only. The presenter walks toward the camera across the set and points at a large screen behind her. She says, "Three taps, and your month is planned." Clean studio sound, no music.
@Image1 is the storyboard: follow its shot order, framing and camera notes. @Image2 is the main character and @Image3 is the forest location. Render it as a warm hand-painted 2D animation with soft grain, gentle wind and birdsong.
@Image1 is the character, @Video1 sets the choreography and timing. The character from @Image1 performs the dance from @Video1 on a rooftop at sunset, outfit and face unchanged. Do not copy the person or background from @Video1. Wind and the city below.
@Image1 is the bakery, @Image2 is the baker. @Audio1 is the narrator's voice-over; generate no other voice. The baker opens the shutters at dawn, pulls trays of bread from the oven and arranges them in the window. Time each action to the narration and end on her flipping the sign to open.
The woman in the photo hangs white sheets on a line in a sunny garden. A breeze lifts the fabric, she smooths it with both hands and turns to smile at the camera. Slow push-in. Birds and fabric flapping.
The camera rises slowly over the mountain lake in the image as mist burns off the water and the first sunlight hits the far peaks. A single canoe drifts into frame from the left. Water lapping, distant loons.
The illustrated fox in the picture blinks, flicks its tail and trots off along the path, leaves scattering behind it. Keep the hand-painted look and colors of the image. Static camera, soft forest ambience.
The product on the table in the image turns slowly on its own as warm light sweeps across it from left to right. Keep the label text sharp and unchanged. Locked-off camera, a soft low hum.
The street in the photo comes alive at night: shop signs flicker on, a scooter passes from right to left and steam rises from a food stall. The camera tracks slowly forward at eye level. Traffic, chatter and sizzling.
by ByteDance
30-second clips in one pass, guided by up to 50 references
From 51 credits per video
by ByteDance
From 45 credits per video
by Google
Video and sound in one pass, 3 to 10 seconds up to 4K, with clip editing
From 14 credits per video
Seedance 2.0 is on the Hobby plan and higher. Create a free workspace to try the other models first, then upgrade when you need it.