How to Use Seedance 2.0
A hands-on walkthrough of the stuff that actually matters: the @tag system, prompt structure, and the mistakes that burn credits for nothing.
Seedance 2.0 Isn't a Text-to-Video Box
Most people treat it like one. They type a paragraph, hit Generate, and wonder why the result looks off. Here's the thing: Seedance 2.0 is a conditioning engine. It takes your text prompt, yes — but it also takes images, video clips, and audio files as explicit references. Each one can anchor a different part of the output: character identity, camera movement, art style, soundtrack tempo.
The difference between a mediocre generation and a usable one usually isn't the prompt. It's whether you told the model what each uploaded file is supposed to do.
Learn the @Tag System (This Is the Whole Game)
You can upload up to 12 files per generation — 9 images, 3 video clips, 3 audio tracks. But uploading files without telling the model their purpose is like handing a camera operator a stack of photos and saying “figure it out.” The @tag system fixes that.
The syntax is simple: @Image1, @Video2, @Audio1 — then describe what role that file plays.
Pin the opening shot
@Image1 as the first frameLocks composition, color palette, and framing from frame one. Useful when you have a specific visual you want to start from.
Keep a face consistent
@Image2 is the main characterThe model tracks this face throughout the clip. Works best with a clear, front-facing headshot — avoid group photos.
Copy camera movement
Reference @Video1's camera tracking and dolly movementThe model replicates the cinematography, not the content. A handheld tracking shot of a dog park can direct a sci-fi corridor chase.
Set the mood with music
@Audio1 as background musicSeedance syncs scene rhythm to the audio beat. Tempo, mood shifts, and drops all influence pacing and cuts.
Transfer artistic style
@Image3 is the art style referenceColor grading, texture, and visual tone get pulled from this image. One reference is enough — stacking multiple style refs muddies the output.
Combine everything
@Image1 as first frame, @Image2 is the main character. Reference @Video1's camera movement. @Audio1 as background music.Each file has a clear job. This is the core idea: be explicit about what each asset does, or the model guesses.
Write Prompts That Actually Work
There's a formula: Subject + Action + Scene + Camera + Style. Stay under 80 words. Shorter, structured prompts beat long descriptions every time — the model doesn't reward verbosity.
Putting It Together
“A woman in her 30s, dark hair pulled back, navy linen blazer, turns slowly toward the camera and smiles. Standing on a rooftop terrace at sunset, city skyline behind her. Medium close-up, slow dolly-in. Soft key light from the left, warm rim light, shallow depth of field, film grain.”
62 words. Five clear anchors. No poetry, no ambiguity.
Pick Your Settings
Not much to overthink here, but a few choices matter more than you'd think.
Duration
4–15 seconds. Start at 4–5 seconds while you dial in the look. Longer clips cost more credits and amplify any problems in your prompt.
Aspect Ratio
6 options: 16:9, 9:16, 4:3, 3:4, 1:1, 21:9. Use 9:16 for Reels/Shorts/TikTok. 16:9 for YouTube and widescreen. 21:9 if you want that ultra-wide cinematic bar look.
Model Variant
Seedance 2.0 Fast — half the credits, 2x speed. Use it for drafts. Seedance 2.0 — full quality. Use it for final renders. Same input capabilities, same lip-sync languages.
Mistakes That Waste Your Credits
You're going to hit these at some point. Knowing them upfront saves you a lot of trial-and-error credits.
Cramming 3+ characters into one scene
What happens: Faces drift, bodies warp, someone grows an extra arm
Fix: Two characters max per scene. Keep clear physical distance between them. If you need a crowd, make them background elements and only describe one or two in detail.
Multiple camera movements in one shot
What happens: Jittery, unstable footage that looks like a broken gimbal
Fix: Pick one move: a slow dolly-in, a gentle pan, or a static lock. Never combine whip pans with zoom or tracking in a single generation.
Writing a 200-word essay as a prompt
What happens: The model picks random details, ignores others, and invents things you didn't ask for
Fix: 30–80 words. Structured: Subject + Action + Scene + Camera + Style. If it doesn't fit in a tweet, it's too long.
Not assigning roles to uploaded files
What happens: Reference images get interpreted randomly — your character ref becomes a background, your style ref becomes a character
Fix: Always use @tags with explicit role descriptions. "@Image1 as first frame" beats uploading an image and hoping.
Expecting readable text in the video
What happens: Signs, titles, and subtitles come out garbled or unreadable
Fix: Either skip on-screen text entirely, or make it large, centered, high-contrast, and single-word. Multi-line paragraphs won't work.
Fast hand gestures or detailed finger work
What happens: Extra fingers, fused hands, impossible anatomy
Fix: Slow down all gestures. "Gently raises one hand" works. "Rapidly gestures while counting on fingers" doesn't.
Tips That Actually Move the Needle
Once you've got the basics down, these are the habits that separate decent results from consistently good ones.
Start short, then extend
Generate a 4–5 second clip first. Lock in the look, the motion, and the character. Then use longer durations or extend the clip. Going straight to 15 seconds with an untested prompt is a credit gamble.
Trim your video references
Don't upload a 15-second clip when you only need 3 seconds of camera movement from it. Trim to the exact segment you want the model to reference. Cleaner input = cleaner output.
One style, one finish
Pick a primary style and one supporting aesthetic. "Wes Anderson color palette with film grain" works. "Wes Anderson meets cyberpunk noir with anime influences" doesn't — the model can't resolve conflicting directions.
Use the Fast variant for exploration
Seedance 2.0 Fast costs half the credits and runs 2x faster. Use it to nail down your prompt and references, then switch to standard Seedance 2.0 for the final render when you want maximum fidelity.
Iterate one variable at a time
Changed the prompt AND swapped a reference image AND adjusted duration? You won't know which change caused the improvement (or regression). Change one thing per generation.
Front-facing headshots for character ref
Character consistency works best with a clear, well-lit, front-facing portrait. Side profiles, group shots, or heavily stylized illustrations give the model less to lock onto.
The 60-Second Workflow
Once you've internalized the @tag system and prompt structure, the actual process is fast.
Prep your assets
Gather a character headshot, a style ref, maybe a video for camera movement. Trim videos to the exact segment you need.
Write a structured prompt
Subject + Action + Scene + Camera + Style. Under 80 words. Assign @tags to every uploaded file.
Draft with Fast
Use Seedance 2.0 Fast for 2–3 quick iterations. Change one variable per run. Lock in the look.
Final render
Switch to standard Seedance 2.0 for the keeper. Set your target duration and aspect ratio. Export.
Common Questions
How long does it take to generate a video with Seedance 2.0?
Depends on duration and reference complexity. A 4-second text-only clip usually takes under a minute. A 15-second clip with multiple image, video, and audio references can take a couple of minutes. Seedance 2.0 Fast roughly halves these times.
Do I need all four modalities to get good results?
No. Text-only works fine for simple scenes. Adding a reference image for the first frame or a character dramatically improves consistency. Audio and video references are optional power tools — use them when you need rhythm sync or specific camera work.
What's the difference between Seedance 2.0 and Seedance 2.0 Fast?
Same capabilities — same 4-modality input, same native audio, same lip-sync languages. Seedance 2.0 Fast runs about 2x faster at half the credit cost, with a slight trade-off in fine detail for complex scenes. Use Fast for drafts and iteration, standard for final renders.
Can I upload a video and just change one element?
Partially. You can use a video as a reference to maintain camera movement, pacing, and composition while changing the subject via your prompt and image references. It's not frame-exact editing, but it's close enough for most creative workflows.
My characters keep changing appearance between shots — how do I fix that?
Use a dedicated character reference image with @tag (e.g., "@Image1 is the main character"). Use a clear, front-facing headshot. Avoid switching reference images between generations if you need consistency across a multi-shot sequence.
Is there a way to extend a clip beyond 15 seconds?
Not in a single generation. The max is 15 seconds. For longer sequences, generate individual clips with consistent character references and edit them together. Starting each new clip from the last frame of the previous one helps with visual continuity.
Showcase
AI-Generated Videos
Discover stunning AI-generated videos. From photography to architecture, people to creative projects — see how AI transforms ideas into cinematic videos.
Theory's Over. Go Make Something.
You know the @tags, the prompt structure, and the mistakes to avoid. The rest is iteration.
Try for Free