Seedance 2.0 Prompt Assistant
Write Seedance 2.0 prompts with a clear order of shots, a job for every reference and motion from frame to frame.
Start from a use case
- Storyboard a 10-second sceneThree shots, each with its own size, camera and one action.Shot plan
- Keep a character from a photoThe person in your picture stays the same from the first step to the last.References
- Animate a still towards an endingMotion that starts on your picture and settles on a closing frame.Frames
- Write a short dialogue beatYour lines kept word for word, with who says them and how.Prompt
- Diagnose a jumpy resultFind out why faces or props jumped between shots.Fix
What you provide
- The moment and its lengthOne scene of 4 to 15 seconds: who, where, what happens and how it ends.
- Which reference does whatUp to nine pictures, three videos and three sounds, each marked as a person, product, place, motion, camera or voice.
- The one thing to protectA face, an outfit, a product or the last frame, so the prompt spends its words there.
How it works
- Read the order of eventsThe clip is built as a sequence, because this model follows the order of actions better than clock times.
- Assign referencesOne line per reference, one person per picture, unused references set aside.
- Write the shotsEach shot gets a size, a camera behaviour and one action that ends in a visible state.
- Hand it to the generatorPrompt, shot list, checks and settings, ready to open with Seedance 2.0.
One idea, before and after
skateboarder does a kickflip down a big set of stairs, slow mo, lands it and his friends cheer
Golden hour sunlight casting long dramatic shadows across a wide concrete public plaza staircase. Shot 1: Medium tracking shot, smooth handheld tracking alongside from the right — A skateboarder in a casual t-shirt and denim approaches the top lip of a large concrete stair set, bends knees, pops the skateboard tail off the pavement, and launches forward into the jump. Shot 2: Low-angle close-up, static perspective from near the base of the stairs — In slow motion, the skateboard rotates in a crisp kickflip directly beneath the rider's sneakers mid-air, levels out, and slams solidly onto the flat ground below as the skater absorbs the impact and rolls forward to screen right. Shot 3: Wide static shot looking from the bottom of the stairs — By the metal handrail on the left, friends raise both arms and pump their fists in celebration as the skater rolls smoothly past them, pumping a fist in response. Sound: The sharp loud snap of a wooden skateboard deck popping, gentle whoosh of air, a thunderous resonant clack of wheels landing flush on concrete, and background vocal cheers and clapping. No on-screen text, no subtitles, no captions, no watermark. Coherent anatomy, natural hands.
Questions
How is this different from the Seedance 2.5 assistant?+
Seedance 2.0 runs 4 to 15 seconds, takes fewer references and follows shot order better than timestamps. This assistant plans for those limits.
Does it write timestamps?+
Yes, as a budget: segments from 0:00 to the end with the order of actions kept unmistakable, because this model follows order more than the clock. Your own times are kept.
How many references fit?+
Up to nine pictures, three videos and three audio clips. Each gets one job, so faces and props stay where they belong.
Does it write dialogue?+
It keeps the lines you wrote, names the speaker and the language, and never invents new lines.
Does it work for image-to-video?+
Yes. With a first frame, and optionally a last frame, it writes only the motion between them and how the camera travels.
What does a run cost?+
A few credits, shown on the button before you run. You never pay more than that price, and nothing is charged if no usable result comes back.