Quick Start: The 3-Minute Video

1

Upload

Drag & drop your reference image (character/style) into the 'Image' tab.

2

Prompt

Describe the action simply: 'A cinematic shot of [reference] walking in rain.'

3

Generate

Hit generate. Wait ~60s. Preview results and download if satisfied.

The Core Workflow

Seedance 2.0 treats video generation like a director treats a shoot. You are not just typing text; you are casting actors (images), setting the scene (prompts), and choosing the soundtrack (audio).

1. Asset Preparation

  • Character Sheet: Use a clear, front-facing photo of your character. Avoid messy backgrounds.
  • Style Reference: (Optional) An image defining the color palette or lighting.

2. Motion Reference

Upload a video clip to guide the movement. Tag it in your prompt to say "Use the camera movement from this video".

Prompting Best Practices

Structure your prompts for maximum control.

The Formula
[Subject Definition] + [Action/Movement] + [Camera Angle] + [Atmosphere/Lighting]
Example:
"A cyberpunk samurai (@image1) unsheathing a katana slowly, low angle dolly zoom, neon rain, volumetric fog, 8k resolution."

Audio Sync

Seedance 2.0 can generate sound from video content, OR generate video from sound.

Lip Sync

Upload a vocal track. The model will animate the reference face to match the speech.

Beat Sync

Upload music. The video cuts and motions will align with the rhythm/beat.

Advanced: Achieving Consistency

The holy grail of AI video is character consistency. Seedance 2.0 solves this with Multi-Reference Tagging.

Strategy 1: The "Face + Outfit" Split

Don't rely on one image. Split your references to guide different aspects of the generation.

Input
  • • @image1: Close-up of face
  • • @image2: Full body shot for outfit
Prompt

"Character with face from @image1 and outfit from @image2 walking down the street..."

Multi-Shot Consistency

To keep a character consistent across different scenes (e.g., Scene A: Living Room, Scene B: Park):

  • Use the exact same @image reference for every generation.
  • Keep the Subject Description in your prompt identical (e.g., "A young woman with red hair").
  • Only change the Action and Environment words.

Motion Transfer

Want specific acting? Upload a video reference.

"Follow the motion and acting from @video1, but replace the actor with the character from @image1."

Common Issues

Faces look distorted in motion

Use a higher resolution reference image. Try to avoid extreme close-ups combined with fast motion.

Text in video is garbled

AI struggles with specific text. Avoid prompts asking for signs/logos unless using post-production overlay.

Generation failed / Timeout

High demand on servers. Try again during off-peak hours or check your credit balance.