Zero Kit AI

Seedance 1.0 AI Video Generator

  • ByteDanceByteDance
  • Up to 1080p
  • Up to 2 reference images
  • Up to 4 videos per run
  • 2–12 seconds
A skateboarder floats mid-ollie over a set of steps in an empty sunlit plaza.

What is Seedance 1.0?

Seedance 1.0 is the first generation of ByteDance's Seedance video models, and it was built for stability. Subjects keep their shape through large movements, from a runner in mid-stride to a chef tossing a wok.

When one prompt asks for several shots, the clip keeps the same main subject, style and atmosphere across the cuts. It works from text alone or from a first-frame image, in styles that range from photorealism to illustration.

Seedance 1.0 Fast is the newer of the two and costs less per second. ByteDance says it builds on the rich detail and material realism of Pro, with more expressive faces and hair.

A good routine is to draft with Fast, compare a take on Pro when a shot matters, and keep whichever look you prefer.

For a stronger film look, Seedance 1.5 Pro is the next step up in the family.

What Seedance 1.0 does well

The same person in every frame

Seedance 1.0 keeps the main subject steady through a clip, including across cuts when one prompt asks for several shots. ByteDance built it to hold the subject, the visual style and the atmosphere through shot transitions.

The cyclist sample shows the goal: the same face, green jacket, grey helmet and bicycle as she rides past a bakery, turns beside a canal and stops at a traffic light to look back.

Describe the character once in full, then repeat the same key details in each shot. Starting from an image of the character as the first frame makes the look even easier to hold.

Three frames of the same cyclist in a green jacket, consistent in face and outfit across the sequence.

A horizontal strip of three cinematic video frames showing the same young cyclist in a green jacket and grey helmet riding through a city: first pedaling past a bakery, then turning a corner beside a canal, then stopping at a traffic light and looking back over her shoulder. Same face, hair, jacket, helmet and bicycle in all three frames, bright daylight, thin black borders.

Fluid movement

Large movements are where Seedance 1.0 does its steadiest work. ByteDance describes a wide dynamic range: from a small change of expression up to a full-body move, the motion stays smooth and physically believable.

A parkour vault like the sample is a good test, with one hand on the wall, the body horizontal in mid-air and the hoodie rippling with the motion.

Write the movement as a sequence, such as "runs, plants one hand, swings both legs over, lands", rather than a single verb. A low side angle keeps the whole arc of the body in frame.

A parkour runner vaults over a concrete wall, body stretched mid-air.

Cinematic 16:9 video keyframe, a parkour runner in a grey hoodie vaulting over a low concrete wall with one hand, body horizontal mid-air, legs extended, captured from a low side angle against a clear sky, fabric rippling with the motion. Smooth, athletic, crisp subject against a softly blurred urban background.

Fast action, stays clear

Fast subjects stay readable in Seedance 1.0 instead of dissolving into blur. For the Fast version, ByteDance points to stable footage in extreme-sports scenes that follows physical laws and avoids distortion and artifacts.

A border collie leaping for a frisbee is a good example: the fur, the flying ears and the kicked-up grass should stay sharp while the background carries the motion blur.

Put the peak of the action in the prompt, name the camera, such as "tracking at a low angle", and keep the background simple so the detail goes into the subject.

A border collie leaps to catch a frisbee on a sunny field.

Cinematic 16:9 video keyframe, a border collie leaping high off a sunny grass field to catch a red frisbee, body fully stretched, ears flying, grass blades kicked up behind it, the camera tracking at a low angle. Sharp fur detail with natural motion blur in the background.

How to write prompts for Seedance 1.0

  1. Start with one subject and one clear action; Seedance 1.0 is steadiest when the movement has a single focus.

  2. Break a big movement into steps, such as take-off, peak and landing, so the model knows the shape of the motion.

  3. Name the camera: "side tracking shot matching her pace", "low angle" or "static wide shot". A camera that follows the subject keeps it sharp.

  4. For several shots in one clip, describe each shot in order and repeat the character's key details in each one.

Sample prompts to try with Seedance 2.5

  • A runner moves along a coastal track at sunrise, the camera tracking alongside.

    Runner at sunrise

    Cinematic 16:9 video keyframe, a side tracking shot of a runner in a light blue top running along a coastal track at sunrise, mid-stride with both feet off the ground, the sea glittering behind, the camera matching her pace so she stays sharp while the railing blurs. Clean bright colors, smooth motion.

  • A girl swings high on a rope swing under an oak tree, her dress fluttering.

    Girl on a swing

    Cinematic 16:9 video keyframe, a girl on a rope swing at the highest point of her arc under a large oak tree, legs stretched out, yellow dress fluttering, leaves falling, sunlight flickering through the canopy. The camera placed low, capturing the arc of motion, playful and fluid.

  • A chef tosses noodles in a flaming wok, the noodles suspended mid-air.

    Wok toss

    Cinematic 16:9 video keyframe, a chef in a white jacket tossing noodles and vegetables high above a flaming wok in a busy kitchen, food suspended mid-air, a burst of orange flame, steam swirling. Fast action frozen crisply, warm kitchen light, dynamic and fluid.

  • An underwater shot of a swimmer gliding past, bubbles trailing behind.

    Swimmer underwater

    Cinematic 16:9 video keyframe, an underwater shot of a swimmer in a red cap gliding through a sunlit pool after a push-off, arms extended forward, a trail of silver bubbles behind, rippling light patterns on the tiled floor. Smooth motion, crystal-clear blue water.

Sample images are illustrative; results vary by model.

Seedance 1.0 limitations

  • Lettering inside the frame is not reliable. Signs, jersey numbers and shop names tend to smear or change between frames.

    Leave on-screen text out of the prompt and add titles in an editor afterwards.

  • Close contact between several people, such as a dance couple or a crowded tackle, is harder than one moving subject.

    Hands and limbs can merge where bodies touch, so check those shots frame by frame before you use them.

  • Keep the subject large in the frame when you work at the free resolution.

Seedance 1.0 FAQ

Seedance 1.0 Fast costs less per second, and ByteDance says it builds on the detail and material realism of Pro with more expressive faces and hair.

Both are versions of the same model, so one prompt works on either.

Ready to try Seedance 2.5?

Write a line or pick a sample prompt, then press generate.

Last updated: