The Complete Guide to
Seedance 2.0 on Loova

Learn how to use Seedance 2.0 on Loova to create high-quality AI videos. This guide explains the core reference system, prompt writing methods, and common workflows so you can generate videos with greater control and consistency.

Seedance 2.0 is one of the most advanced AI video generation models available today. Instead of relying only on text prompts, it allows you to combine text, images, videos, and audio as references. This makes it possible to guide the model more precisely and generate videos closer to your intended result.

With Seedance 2.0 on Loova, you can:

• Generate videos from text prompts
• Use images or videos as references
• Replicate camera movement and choreography
• Maintain visual consistency across scenes
• Extend or edit existing videos
• Add voice, dialogue, and environmental sound

This guide walks through the main features and practical ways to use them.

Table of Contents

• General Questions
• The Multi-Modal @ Reference System
• Text-to-Video Generation
• Maintaining Visual Consistency
• Camera Movement & Motion Reference
• Style & Effect Replication
• AI Story Expansion
• Video Extension
• Audio, Voice & Lip Sync
• Continuous One-Take Shots
• Video Editing
• Prompt Writing Basics
• Cinematography Language for Prompts
• Character Consistency Tips
• Video Quality Control
• Multi-Shot Storytelling
• Practical Use Cases
• Pro Tips
• Common Mistakes to Avoid
• Prompt Templates

General Questions

1. Long Generation Times

Seedance 2.0 is currently experiencing unusually high demand across the entire ecosystem, including the official model platform. During peak hours, this may result in longer processing times or queue delays.

Our engineering team is actively expanding generation capacity to improve stability and reduce wait times as quickly as possible. The current infrastructure upgrade is expected to be completed between mid-March and the end of March.

In the meantime, we recommend using Seedance 2.0 Fast, which is optimized for handling peak traffic and typically completes generations much faster. Many users switch to this mode during high-load periods for a smoother experience.

We appreciate your patience while we continue improving the system capacity.

2. Generation Fails or Doesn’t Complete

For Seedance 2.0, the model currently applies strict content moderation rules at the model level. Based on information from the official provider, certain prompts or visual inputs may trigger automatic restrictions. This may include:

• recognizable characters

• specific faces or identities

• certain sensitive visual contexts

To help users work more effectively with the model, we’ve prepared this Seedance 2.0 prompt guide in the following text that summarizes the known limitations and recommended prompt structures.

Please double-check your prompt and any uploaded images or videos to ensure they don’t include any of the restricted content listed above.

If you’d like, you can also share your prompt and inputs with us, and we will gladly review them to help identify the issue more quickly.

Multi-Modal @ Reference System

The @ reference system is a key feature of Seedance 2.0 on Loova. It allows you to clearly tell the model how each uploaded file should be used in your video generation.

Seedance 2.0 supports multiple input types, including text, images, videos, and audio. These materials can be used as references for different parts of your video, such as:

• Characters or subjects
• Motion and actions
• Visual style or effects
• Camera movement
• Scenes or environments
• Background music and sound effects

As long as your prompt clearly explains how each asset should be used, the model will follow your instructions.

How the @ System Works

When you upload files in Loova, Seedance 2.0 automatically assigns labels to them, such as:

• @Image1, @Image2
• @Video1, @Video2
• @Audio1

You can then mention these labels directly in your prompt to define the role of each file in the generated video.

Example prompt

@Image1 as the first frame, reference @Video1 for camera movement, and use @Audio1 as the background music.

Instead of only describing a scene, the @ system lets you assign specific roles to each file, giving you much more precise control over the final result.

Basic Workflow

Using the @ reference system is simple:

1. Upload your assets Add images, videos, or audio files in Loova.
2. Automatic labeling The system assigns labels like @Image1, @Video1, or @Audio1.
3. Reference them in your prompt Tell the model exactly how each file should be used.
4. Generate the video Seedance 2.0 combines your instructions and assets to create the final video.

Common Usage Patterns

The multi-modal system supports many creative workflows. Here are several useful patterns.

3. First Frame + Motion Reference

You can define the starting image while borrowing motion from another video.

Example:

@Image1 as the first frame, reference @Video1’s fighting choreography.

This keeps the visual appearance from the image while applying motion from the video.

4. Extend an Existing Video

You can continue a video by generating additional seconds.

Example:

Extend @Video1 by 5 seconds.

**Important:**Set the generation duration to only the added portion. For example, if you extend by 5 seconds, the generation length should also be 5 seconds.

3. Insert a Scene Between Videos

You can generate a new segment between two existing clips.

Example:

Add a scene between @Video1 and @Video2. The content is xxx.

This allows you to create smooth transitions or additional story moments.

4. Use Audio From a Video

You can reuse audio directly from a referenced video.

Example:

Reference @Video1’s background music and sound effects.

No separate audio upload is required.

5. Create Continuous Character Actions

Multiple images can be used to guide a sequence of actions.

Example:

The character transitions from a jump directly into a roll, maintaining smooth motion. @Image1 @Image2 @Image3

Seedance will interpret the sequence and generate fluid movement between the frames.

Important Tips

When working with multiple assets, keep these points in mind:

• Check that each @ label is correct (@Image, @Video, @Audio).
• Clearly explain the role of every referenced file.
• Avoid mixing up assets (for example, using a video as a character reference without specifying it).

Clear prompts help Seedance 2.0 better understand your intent and produce more accurate results.

Ultra-Realistic Text-to-Video

With Seedance 2.0 , you can generate high-quality videos using only text prompts, even without uploading any reference images, videos, or audio.

The model is designed to understand natural language descriptions and can handle complex, multi-step actions in a single prompt. As long as the scene and actions are described clearly, Seedance 2.0 can generate smooth motion, believable behavior, and coherent storytelling.

This makes it easy to quickly turn ideas, concepts, or short scripts into visual scenes.

Example Prompts

Below are several examples showing how detailed text prompts can guide the model to generate realistic and cinematic results.

Everyday Realism — Laundry Scene

A simple daily activity with natural motion.

Example prompt

A girl elegantly hanging clothes to dry. After hanging one piece, she reaches into the bucket for another, gives it a vigorous shake, and hangs it up.

This type of prompt focuses on ordinary actions performed step-by-step, allowing the model to generate smooth and realistic movement.

Painting Comes to Life — Commercial Concept

A creative concept suitable for advertising or storytelling.

Example prompt

The person in the painting looks nervous, eyes darting left and right, then peeks out from the frame. They quickly reach out of the painting to grab a cola, take a sip, and show a satisfied expression. Footsteps approach — the person in the painting hurriedly puts the cola back. A cowboy walks up and takes the cola, then leaves. The camera pushes in as the screen gradually fades to black with only a top light illuminating a can of cola. Artistic subtitles appear at the bottom: “Yikou Cola — a taste you can’t miss!”

This example demonstrates how a multi-scene narrative can be described directly through text.

Period Drama — 19th Century London

Historical environments and cinematic camera movement.

Example prompt

Camera slowly pulls out, revealing the full street, and follows the female lead. Wind blows her skirt as she walks along a 19th-century London street. A steam-powered car speeds past from the right side, its wind lifting her skirt. She gasps in shock and quickly presses her skirt down with both hands. Background audio includes footsteps, crowd chatter, and vehicle sounds.

Here the prompt describes camera movement, environment, character reactions, and ambient audio.

Action Sequence — Chase Scene

Fast-paced motion with dynamic camera work.

Example prompt

Camera follows a man in black sprinting in a desperate escape, with a crowd chasing behind him. The camera shifts to a side tracking shot. He panics, crashes into a roadside fruit stand, quickly scrambles up, and continues running. The sound of a chaotic crowd fills the background.

This prompt shows how Seedance 2.0 can interpret continuous action sequences and camera changes from natural language.

Tips for Writing Text-Only Prompts

To get better results when generating videos from text:

• Describe actions step by step rather than all at once.
• Include camera movement when needed (e.g., follow shot, push-in, wide reveal).
• Add environmental details such as location, time period, or weather.
• Mention character reactions and body movement to make scenes more natural.
• Include sound or atmosphere if it’s important to the scene.

Clear and structured prompts help Seedance 2.0 better understand your idea and produce more coherent videos.

Consistency Across Every Detail

One of the most common challenges in AI video generation is visual consistency. In many models, faces may change during a clip, product details can disappear, text may become distorted, or the visual style may shift unexpectedly.

Seedance 2.0 on Loova addresses this issue by maintaining strong consistency throughout the entire video generation. From character identity to product details and visual style, the model keeps key elements stable from beginning to end.

This means the generated video remains coherent, recognizable, and visually stable, even in longer or more complex scenes.

What Stays Consistent

Seedance 2.0 is designed to preserve multiple aspects of the scene, including:

• Face identity — characters remain recognizable throughout the video
• Product details — logos, text, and small visual elements stay intact
• Environment — locations remain stable across shots
• Visual style — no unexpected style shifts during the clip
• Clothing and accessories — outfits and items remain consistent

This helps ensure that characters, products, and scenes stay visually aligned across the entire generation.

Example Prompts

Below are several examples showing how Seedance 2.0 maintains consistency across different scenarios.

Multi-Scene Family Reunion — Character ​Consistency ​​ in a Narrative

Example prompt

The man (@Image1) tiredly walks down a hallway after work, his steps slowing. He stops at the front door. Close-up on his face — he takes a deep breath, adjusts his emotions, releases the negativity, and relaxes. Close-up of him finding his keys, inserting into the lock. After entering, his young daughter and a pet dog joyfully run over to greet him with hugs. The interior is warm and cozy. Natural dialogue throughout.

In this example, the same character from @Image1 remains consistent across multiple shots and emotional moments.

Style Transformation While Keeping the Character

Example prompt

Replace the girl in @Video1 with a Chinese opera huadan. The scene is an ornate stage. Reference @Video1’s camera work and transitions. Use the camera to match the character’s movements. Ultimate stage aesthetics with enhanced visual impact.

Here, the character identity remains stable, while the visual style is transformed into a Chinese opera stage aesthetic.

One Continuous Take With Scene Changes

Example prompt

Reference all transitions and camera work from @Video1. One continuous take. The scene opens on a chess game, camera pans left revealing yellow sand on the floor, camera tilts up to a beach with footprints. A girl in white walks into the distance on the beach. Camera cuts to an aerial overhead view — ocean waves washing the shore (no people visible). Seamless dissolve transition — the washing waves transform into flowing curtains. Camera pulls back to reveal the girl’s face in close-up. One continuous take throughout.

This prompt demonstrates how consistent characters and environments can be maintained even as the scene evolves.

Product Commercial — Maintaining Brand Details

Example prompt

Perform a commercial-grade showcase of the handbag in @Image2. The side view references @Image1. The surface material references @Image3. Show all details of the handbag. Background audio should be grand and majestic.

Seedance 2.0 keeps product features, textures, and design details consistent across the video.

Multi-Language Product Commercial

Example prompt

0–2s: Rapid four-frame flash cuts — red, pink, purple, and leopard-print bows displayed in sequence. Close-ups of satin sheen and the “chéri” brand text. 3–6s: Close-up of silver magnetic clasp clicking shut, then gently pulled apart — showcasing silky texture and convenience. 7–12s: Quick cuts of wearing scenarios — wine-red bow on a coat collar for commuter style; pink bow tying a ponytail for sweet street look; purple bow on a bag strap for understated luxury; leopard-print bow on a suit collar for bold style. 13–15s: All four bows displayed side by side with brand name.

This type of prompt demonstrates how brand elements and product visuals remain clear and stable across multiple scenes.

First-Person POV With Multi-Scene References

Example prompt

Set @Image1 as the first frame. First-person perspective. Reference @Video1 for camera movement. Upper scene references @Image2, left scene references @Image3, right scene references @Image4.

Here, multiple references help maintain scene layout and perspective consistency within a first-person viewpoint.

Tip for Better Consistency

To improve results when generating videos:

• Clearly specify which asset defines the main subject or character
• Use reference assets (images or videos) when identity matters
• Describe scene transitions and camera movement clearly
• Keep prompts structured when working with multiple shots

This helps Seedance 2.0 better maintain visual stability and narrative continuity throughout the video generation process.

Camera & Motion Replication

In the past, recreating specific film techniques or motion styles with AI video required complicated prompts, and the results were often unreliable. With Seedance 2.0 on Loova, this process becomes much simpler.

Instead of describing everything in detail, you can upload a reference video, and the model will analyze and replicate key elements such as motion patterns, camera techniques, and editing rhythm.

Seedance 2.0 can replicate:

• Choreography — fight scenes, dance moves, action sequences
• Camera techniques — dolly shots, tracking shots, crane movements, handheld motion, Hitchcock zoom
• Editing rhythm — timing of cuts, transitions, and pacing
• Specific cinematic moves — whip pans, orbit shots, mechanical-arm tracking

By referencing a video, the model can extract the motion language and recreate it in new scenes.

Example Prompts

Below are several examples demonstrating how reference videos can guide camera motion and character actions.

Hitchcock Zoom + Mechanical Arm Tracking

Example prompt

Reference @Image1 for the man’s appearance. He is in @Image2’s elevator. Fully replicate @Video1’s camera movements and the protagonist’s facial expressions. Use a Hitchcock zoom when he becomes startled, followed by several orbit shots showcasing the elevator interior. When the elevator doors open, a tracking shot follows him walking out. The exterior scene references @Image3. The man looks around. Reference @Video1’s mechanical-arm multi-angle tracking to follow the character’s line of sight.

This prompt uses a reference video to reproduce specific cinematic camera techniques.

Complex Chase Sequence With Multiple Camera Techniques

Example prompt

Reference @Image1 for the man’s appearance. He is in @Image2’s corridor. Fully replicate @Video1’s camera work and the protagonist’s facial expressions. The camera follows the protagonist sprinting around a corner in @Image2, then into @Image3’s long hallway. The camera tracks from behind at a low angle, orbiting around to the front. The camera pans right 90 degrees to capture @Image4’s forked intersection, stops abruptly, then pans 180 degrees for a close-up of the protagonist gasping for breath. The camera follows the protagonist’s POV looking around — reference @Video1’s rapid left-right orbital camera work to reveal the scene. Pull back to @Image5’s environment and continue side-angle tracking as the protagonist runs.

This example combines dynamic camera movement with complex character action.

Product Showcase With Cinematic Camera

Example prompt

The tablet from @Image1 as the main subject. Camera movement references @Video1 — push in for a screen close-up, then rotate as the tablet flips to reveal its full form. Data streams constantly change on the screen. The surrounding environment gradually transforms into a sci-fi data space.

Here the reference video helps reproduce a cinematic product presentation style.

Dance With Synchronized Camera Rhythm

Example prompt

The female star from @Image1 as the main subject. Reference @Video1’s camera techniques for rhythmic push, pull, and pan movements. The star’s dance also references the choreography from @Video1’s dancer, performing energetically on stage.

Both dance choreography and camera rhythm come from the reference video.

Multi-Character Fight Choreography

Example prompt

Reference @Image1 and @Image2 for the spear-wielding character. Reference @Image3 and @Image4 for the dual-blade character. Replicate @Video1’s choreography. The fight takes place in @Image5’s maple leaf forest.

This prompt recreates complex fight choreography between multiple characters.

Fight Scene With Combined Camera Language

Example prompt

Reference @Video1 for character actions. Reference @Video2 for orbiting camera language. Generate a fight scene between Character 1 and Character 2 under a starry sky. White dust rises during combat. The fight is visually dramatic and the atmosphere intensely tense.

Multiple reference videos can be used to guide both action and camera style.

Commercial Car Showcase

Example prompt

Reference @Video1’s camera work, transitions, and editing rhythm. Replicate the style using the red supercar from @Image1.

This allows you to recreate the cinematic structure of an advertisement with a new product.

Creative Template & Effects Replication

Seedance 2.0 can also analyze visual structure, editing patterns, and effects from a reference video and reproduce them in a new scene.

This means you can take inspiration from a specific video style, editing format, or visual effect, and generate new content with the same structure.

You don’t need to use professional filmmaking terminology. Simply describe what you want to reference.

For example:

Reference @Video1’s rhythm and camera work. Use @Image1 as the character design.

The model will interpret the reference and recreate the style in the generated video.

Example Prompts

VR / Sci-Fi Scene Transition

Example prompt

Replace @Video1’s subject with @Image1. Use @Image1 as the first frame. The character puts on a virtual sci-fi headset. Reference @Video1’s camera work. The perspective changes from third-person view to the character’s POV. Through the AI headset, the scene transitions to @Image2’s deep blue universe where several spaceships appear and fly into the distance. The camera follows the ships, transitioning to @Image3’s pixel world. The camera flies low over pixel mountains with stylized trees. The view tilts upward and accelerates toward @Image4’s green-textured planet, gliding across its surface.

Fashion Montage With Fish-Eye Effect

Example prompt

Reference @Image1 for the model’s facial features. The model wears outfits from @Image2 to @Image6, approaching the camera with playful, cool, cute, surprised, and confident poses. Each outfit change triggers a scene cut. Reference @Video1’s fish-eye lens effect and ghosting flicker visual effects.

Advertisement Recreation

Example prompt

Reference @Video1’s advertisement concept. Use the provided down jacket image, referencing the goose feather image and swan image. Pair with the ad copy: “This is goose down. This is a warm swan. This is a wearable arctic swan-down jacket. Dress warm for the new year, live warm every day.” Generate a new down jacket commercial.

Ink-Wash Style With Effects Transfer

Example prompt

Black-and-white ink-wash style. The character from @Image1 references @Video1’s movements and effects while performing a tai chi martial arts sequence in ink-wash style.

Particle & Texture Effects Transfer

Example prompt

Start from a black screen. Reference @Video1’s particle effects and materials. Gold-gilded sand drifts in from the left side of the frame, gradually covering the scene. Reference @Video1’s particle dispersion effect. The text from @Image1 slowly appears at the center of the frame.

Character Effects Mashup

Example prompt

The character from @Image1 references @Video1’s actions and expression changes while demonstrating the exaggerated act of eating instant noodles.

Puzzle-Shatter Transition Recreation

Example prompt

Start from @Image1’s ceiling. Reference @Video1’s puzzle-shattering transition effect. Replace the “BELIEVE” text with “Seedance.” Reference @Image2’s font style.

AI Creativity & Story Completion

Seedance 2.0 can do more than simply follow prompts. When given a basic idea or starting point, the model can intelligently expand the scene by adding narrative flow, emotion, and visual details.

This allows creators to begin with a simple concept—such as a storyboard, comic panel, or set of images—and let the model help complete the story.

By combining prompts with reference images or videos, Seedance 2.0 can generate content that feels more cinematic and narratively connected.

Example Prompts

Comic Panel to Animated Story

Example prompt

Animate @Image1 as a comic strip, reading left to right and top to bottom. Keep the characters’ dialogue consistent with the text shown in the image. Add sound effects for scene transitions and key moments in the story. The overall tone should be humorous and playful. Reference @Video1 for the animation style.

This prompt turns a static comic image into an animated story sequence.

Storyboard to Cinematic Opening

Example prompt

Reference @Image1’s storyboard. Follow its shot composition, framing, camera movement, visuals, and copy. Create a 15-second calming opening sequence for a piece titled “Childhood Through the Four Seasons.”

A storyboard can guide the structure, while the model fills in visual transitions and cinematic pacing.

Mood-Driven Video From Image Inspiration

Example prompt

Using @Image1 through @Image5 as visual inspiration, create an emotionally driven video. The background music should reference the audio from @Video1.

In this case, the images serve as creative inspiration, while the model builds a cohesive mood and narrative.

Video Extension & Continuation

Seedance 2.0 also supports extending existing videos. Instead of simply generating unrelated footage, the model continues the scene with smooth visual and narrative continuity.

This allows you to extend a clip forward (continuing the story) or backward (adding a prequel).

When extending a video, set the generation duration to match the length of the new segment. For example, if you want to extend a clip by 5 seconds, set the generation duration to 5 seconds.

Example Prompts

Commercial Extension — Donkey Motorcycle Ad (15s)

Example prompt

Extend the video by 15 seconds. Reference @Image1 and @Image2’s donkey-on-motorcycle character. Scene 1: Side fixed shot — the donkey bursts out of a fence while riding a motorcycle, startling nearby chickens. Scene 2: The donkey performs donuts on sand. Close-up of the motorcycle tire, then an aerial overhead shot of the donkey spinning and kicking up dust. Scene 3: Snowy mountain background — the donkey rides over a hillside on the motorcycle. The ad text “Inspire Creativity, Enrich Life” appears behind the subject through a masking effect as the motorcycle flies past, ending with a trail of dust.

Fitness Advertisement Extension (6s)

Example prompt

Extend the video by 6 seconds. An intense electric guitar riff begins. The text “JUST DO IT” appears at the center of the screen and gradually fades. The camera tilts upward to the ceiling where a muscular man performs a pull-up on gymnastic rings. He wears @Image1’s tight-fitting workout gear with the @Image2 “Fitness” logo on the back. As he completes the pull-up, the closing text “DO SOME SPORT” appears.

Coffee Commercial Extension (15s)

Example prompt

Extend @Video1 by 15 seconds. 1–5s: Light and shadows slowly move across a wooden table and a cup through venetian blinds, with tree branches gently swaying outside. 6–10s: A single coffee bean slowly falls from above. The camera pushes closer until the bean fills the frame and the screen fades to black. 11–15s: Text fades in: First line — “Lucky Coffee” Second line — “Breakfast” Third line — “AM 7:00–10:00”

Backward Extension — Adding a Prequel (10s)

Example prompt

Extend the video backward by 10 seconds. In warm afternoon sunlight, the camera starts on awnings fluttering at a street corner. It slowly tilts down to small daisies growing near the base of a wall. Red sneakers enter the frame — the protagonist crouches at a flower stand, smiling as he gathers a large bouquet of sunflowers. The petals brush against his white T-shirt. As he turns to step onto his skateboard, the flower stand owner laughs and calls out, “Watch out for flying petals!” The protagonist waves back and begins skating away. Several golden petals fall from the bouquet and land on the skateboard deck.

Audio, Voice & Lip Sync

Seedance 2.0 introduces a powerful audio generation system that synchronizes sound with visuals automatically. Instead of adding audio in post-production, the model can generate voices, sound effects, ambient sound, and music directly during video creation.

The system supports multi-language speech, realistic lip-sync, and context-aware sound design based on what appears on screen.

Key Capabilities

• Accurate lip-sync that matches mouth movements
• Multi-language voice generation (Mandarin, English, Spanish, Korean, and more)
• Environmental sound generation such as wind, rain, and traffic
• Action-based sound effects that match on-screen movement
• Background music generated to match the rhythm of the visuals
• Voice timbre reference from uploaded videos

Example Prompts

Animal Dialogue with Voice Reference

Example prompt

Fixed camera. Central fish-eye lens view looking down through a circular opening. Reference @Video1’s fish-eye lens. Make the horse from @Video2 look into the fish-eye camera. Reference @Video1’s talking movements. Background music references @Video3’s sound effects.

This setup creates dialogue with synchronized mouth movement while using reference audio and visual style.

Real Estate Documentary Voiceover

Example prompt

Using the provided office building photos, generate a 15-second cinematic documentary in a realistic style. 2.35:1 widescreen, 24fps with delicate visuals. The narrator’s voice timbre references @Video1. Show the ecosystem of the office building — different companies operating inside — while the narration explains how the building has become a vibrant commercial hub.

This produces documentary-style narration with a referenced voice tone.

Character Dialogue — Comedy Talk Show

Example prompt

A comedy roast between a cat and a dog in a talk show titled “Cat & Dog Roast Show”. Cat host (licking paw, rolling eyes): “Folks, who even understands? This guy next to me does nothing all day but wag his tail, destroy the couch, and use those ‘please pet me’ eyes to scam treats from humans. And he still goes by ‘Lucky’? I’d call him ‘Wrecky’!” Dog host (tilting head, wagging tail): “You’re one to talk! You sleep eighteen hours a day. The moment you’re awake you rub against humans for canned food. You shed so much their black clothes are completely covered in fur!”

This example demonstrates multi-character dialogue with expressive delivery.

Chinese Opera Dialogue

Example prompt

The prelude to “The Case of Chen Shimei”. On the left stands Judge Bao in black robes pointing toward Chen Shimei in red. In traditional opera singing style he declares: “The blade meets its sheath — the evidence is clear. Do you still dare deny it?” Chen Shimei looks around nervously, embarrassed and searching for escape. Suddenly a female opera voice from off-screen calls: “Hold!” Both characters turn toward the sound.

This prompt demonstrates stylized performance dialogue and dramatic vocal tone.

Multilingual Family Dialogue Scene

Example prompt

Center frame, the girl in the hat happily sings: “I’m so proud of my family!” She turns and hugs the girl next to her. The girl responds emotionally: “My sweetie, you’re the heart of our family.” The boy in yellow on the left shouts: “Folks, let’s dance to celebrate!” The girl on the far right replies: “I’ll bring the music!” Latin music begins. People stomp their feet and children clap along as the entire family forms a circle and dances joyfully on the colorful street.

This demonstrates multiple characters speaking with music and crowd ambience.

Morning Scene with Voice Reference

Example prompt

0–3s: Alarm clock rings. Scene fades in slightly blurred. 3–10s: Quick camera pan to a close-up of the man’s face. He tiredly calls the girl to wake up. His voice and tone reference @Video1. 10–12s: The girl pouts and hides under the blanket. 12–15s: Cut to a full-body shot of the man. He sighs and says: “I really can’t do anything with you!”

Educational Narration

Example prompt

In the tone of a science documentary, animate the content from @Image1. The narration explains the story: Wukong needs the Banana Leaf Fan to cross Flaming Mountain and travels to Emerald Cloud Mountain to borrow it from Princess Iron Fan. She refuses because Wukong previously subdued her son Red Boy and sent him to become Guanyin’s disciple. The conflict escalates as Wukong’s request fails and a confrontation begins.

This format works well for educational storytelling and historical explanations.

One-Take Continuity

Seedance 2.0 is capable of generating long continuous shots without cuts, maintaining consistent characters, environments, and motion throughout the entire sequence.

This style—often called a “oner” shot—is commonly used in cinematic filmmaking.

Example Prompts

Urban Continuous Tracking Shot

Example prompt

@Image1 through @Image5 — one continuous tracking shot following a runner up a staircase, through interior corridors, and onto the rooftop. The sequence ends with an overhead view of the city skyline.

Creative Transformation — Airplane to Ice Cream

Example prompt

Start with @Image1 as the first frame. The camera slowly zooms into the airplane window. Clouds drift across the frame, one of them covered with colorful candy sprinkles. The cloud gradually morphs into @Image2’s ice cream. The camera pulls back into the cabin where the girl from @Image3 reaches out the window, grabs the ice cream, takes a bite, and smiles sweetly. Voiceover references @Video1.

Spy Thriller Continuous Scene

Example prompt

Spy thriller style. @Image1 as the first frame. Camera tracks in front of a female spy wearing a red trench coat as she walks forward. Passersby occasionally block the view. She reaches a corner (building reference @Image2). The camera remains fixed as she turns the corner and disappears. A masked girl (appearance reference @Image3) emerges and stares after her. The camera then moves forward as the spy enters a mansion (@Image4) and disappears. No cuts — one continuous shot.

Cozy Cabin POV

Example prompt

Start from @Image1’s exterior view. First-person POV pushing quickly into the cabin interior. A deer (@Image2) and a sheep (@Image3) sit near the fireplace drinking tea and chatting. The camera pushes closer to the teacup — its design references @Image4.

Roller Coaster POV

Example prompt

@Image1 through @Image5 — first-person POV of a roller coaster ride in one continuous shot. The speed gradually increases as the track twists and drops.

Video Editing

Seedance 2.0 also supports targeted editing of existing videos. Instead of regenerating an entire clip, you can modify specific elements while keeping everything else intact.

Common edits include:

• Character replacement
• Hairstyle or clothing changes
• Adding or removing objects
• Rewriting the narrative while preserving the scene

Example Prompts

Narrative Reversal

Example prompt

Rewrite @Video1’s entire storyline. 0–3s: A man in a suit sits calmly at a bar, slowly swirling a glass. Dramatic lighting. He whispers: “This deal… it’s big.” 3–6s: A woman behind him asks nervously: “How big?” 6–9s: Suddenly he pulls out an absurdly oversized snack gift pack from under the table and slams it down. 9–12s: The woman’s tense posture relaxes as the atmosphere turns playful. 13–15s: Wide shot of the bar. Text appears: “No matter how busy, remember to have a snack.”

Character Swap

Example prompt

Replace the female lead singer in @Video1 with the male vocalist from @Image1. All movements and actions should replicate the original video exactly. Continuous performance with no cuts.

Simple Element Addition

Example prompt

Change the woman’s hairstyle in @Video1 to long red hair. A great white shark from @Image1 slowly rises behind her, with half its head visible above the water.

Commercial Brand Insertion

Example prompt

@Video1: The camera pans right. A fried chicken shop owner hands orders to waiting customers while saying in Mandarin: “After his order, yours — please line up.” He grabs a paper bag printed with @Image1’s brand design. Close-up of the bag and the hand-off to the customer.

Beat-Synced Editing

Seedance 2.0 can also generate music-video style edits synchronized with rhythm. By referencing a video containing music, the model can align cuts, transitions, and motion to the beat.

Example Prompts

Fashion Beat Sync

Example prompt

The girl in the poster repeatedly changes outfits. Clothing styles reference @Image1 and @Image2. She carries the bag from @Image3. The editing rhythm and beat reference @Video1.

Multi-Image Rhythm Montage

Example prompt

@Image1 through @Image7 — characters appear according to @Video1’s keyframe positions and overall rhythm. The visuals should feel dreamlike with strong dramatic tension. Adjust framing of the reference images to match the music and scene transitions. Add lighting changes to enhance the mood.

Landscape Rhythm Montage

Example prompt

@Image1 through @Image6 scenic landscapes. Reference @Video1’s visual rhythm so scene transitions match the musical beat.

Anime Combat with Precise Timing

Example prompt

8-second anime battle clip with a revenge theme. 0–3s: Female lead turns and sits, placing a chess piece while saying “You lose.” (Storyboard 1–2) 3–4s: Close-up of the opponent’s face. He grits his teeth in anger. (Storyboard 3) 4–6s: Overhead shot — she places another piece. Spectators gasp. (Storyboard 4) 6–8s: Camera whips downward. Screen fades to black, then reveals her looking out a window at moonlight saying quietly: “We’ll see about that.” (Storyboard 5)

Emotion Performance

Seedance 2.0 on Loova has greatly improved how characters express emotion in generated videos. Characters can now communicate feelings more naturally through facial expressions, body language, and voice, allowing scenes to feel more believable and dramatic.

Whether you’re creating storytelling scenes, comedic moments, or advertising content, emotional performance helps make the video feel more alive.

Example Prompts

Emotional breakdown

The woman from @Image1 walks toward a mirror. She looks at her reflection — her pose references ​**@Image2​. After pausing for a moment in silence, she suddenly breaks down and screams. Her action of grabbing the mirror and her emotional expression closely follow the emotion and intensity shown in ​@Video1**​.

Comedic transformation

Use @Image1 as the opening frame. The camera slowly rotates and pushes closer to the character. The character suddenly looks up — facial appearance references @Image2 — and lets out a loud roar. The expression is intense but comedic, referencing the mood and demeanor in ​**@Image3​. Then the character’s body transforms into a bear, following the design in ​@Image4**​.

Advertisement with contrasting emotions

This is a ​range hood commercial​. Start with @Image1 — a woman calmly cooking with no smoke around her. The camera quickly pans right to ​**@Image2​, where a man is cooking in heavy smoke, sweating and looking uncomfortable. The camera then pans back left and pushes in toward the range hood placed on the table in ​@Image1​. The range hood (referencing ​@Image4**​) rapidly pulls in all the smoke.

Prompt Writing Fundamentals

A simple structure can help you write more effective prompts for Seedance 2.0:

Subject + Action + Setting + Lighting + Camera + Style + Quality + Constraints

Example

A young woman slowly walking along a beach.

A gentle breeze moves her hair as she smiles toward the camera.

Warm golden-hour lighting.

4K high-definition, cinematic atmosphere, stable camera movement, smooth and natural motion, sharp visual details.

When writing prompts, try to describe the scene in clear, natural language. If you are referencing files, make sure it is clear whether the asset is used as a reference or modification. If multiple assets are included, double-check that each @ label matches the correct file.

Action Description Tips

Seedance 2.0 performs best when actions are described clearly.

Recommended wording

Use motion descriptions that feel natural and continuous:

• slow
• gentle
• smooth
• natural
• fluid

Examples:

• slowly turns around
• gently raises a hand
• light footsteps
• slightly lowers the head
• swaying in the wind

Avoid

Try to avoid prompts that include:

• exaggerated or extreme movements
• very fast actions
• complex multi-character interactions
• unnatural body twisting

Also avoid vague descriptions such as “dancing” or “walking.” Instead, describe how the movement happens.

Camera & Cinematography Language

Seedance 2.0 understands common cinematic camera directions, so you can include them directly in your prompts.

Shot types

• Close-up
• Medium shot
• Wide shot
• Extreme close-up

Camera movements

• Slow push-in
• Gentle pull-out
• Steady pan
• Half-orbit

You can also specify stability and style, such as:

• fixed camera
• stable handheld
• no shake
• very smooth motion

Example

Medium shot with a slow push-in.

The camera steadily follows the character.

Movement is smooth with no shaking.

Character Consistency Tips

When generating videos with the same character, it’s important to keep their appearance stable.

Add constraint phrases like:

• clear facial features
• stable face, no distortion
• natural body proportions
• no deformation
• consistent clothing and hairstyle

Example

The character’s face remains stable with no distortion.

Body proportions stay natural, and movement is smooth and realistic.

Video Quality Control

Adding quality-related keywords helps maintain visual clarity.

Resolution and detail

• 4K
• ultra-high definition
• rich visual details
• sharp resolution

Visual style

• cinematic look
• natural colors
• soft lighting

Stability

• no blur
• no ghosting
• no flickering
• stable footage

Multi-Shot Storytelling

Seedance 2.0 can understand prompts that describe multiple shots within a single story. By writing your shots in order, the model can generate a continuous sequence with smooth transitions.

Example

Shot 1: A wide shot of the city skyline. Shot 2: A close-up of the character from @Image1. Shot 3: An over-the-shoulder tracking shot as the character walks into a building.

The model will generate a scene with consistent characters, synchronized audio, and natural shot transitions.

You can also describe transitions in a more narrative way:

The scene begins with a close-up of a face.

The camera slowly pulls out to reveal the surroundings.

The character starts walking while the camera follows.

The scene ends with a freeze-frame smile.

Real-World Use Cases

Seedance 2.0 on Loova can be used for a wide range of creative and production scenarios.

Ads & E-commerce

• product demonstration videos with narration
• recreating successful ad styles with your own products
• scalable ad templates for marketing campaigns
• brand-consistent multi-scene commercials
• multilingual ads with lip-sync

Content Localization

Generate different language versions of the same video.

• keep the original motion from a reference video
• generate new dialogue in another language
• produce localized versions with matching lip-sync

Short-Form Content

Create social media videos quickly.

• combine clips, audio, and effects
• produce TikTok, Reels, and YouTube Shorts
• generate beat-synchronized videos from still images

Storyboard to Video

Upload storyboard images as references and describe how the scenes should move.

Seedance 2.0 can automatically fill in the motion and transitions between frames.

Tutorials & Educational Content

Create instructional videos such as:

• step-by-step tutorials with voiceover
• AI avatar guides
• animated science explanations

Film & Narrative Scenes

Seedance 2.0 can also support storytelling content such as:

• action sequences with multiple camera angles
• emotional dialogue scenes with lip-sync
• long continuous shots with scene transitions
• historical or period-style settings

Video Remix Ideas

Here are some common workflows you can try when creating videos.

Fashion runway swap

Input: runway video + clothing image

Prompt example:

Replace the clothing worn by the model in @Video1 with the outfit shown in ​**@Image1**​.

Element addition or removal

Input: video

Examples:

Add a glowing neon sign in the background.

Remove the car from the scene.

Character replacement

Input: video + character image

Prompt example:

Replace the main character in @Video1 with ​**@Image1**​, while keeping all original actions.

Video continuation

Input: video

Example:

Extend @Video1 by 10 seconds and reveal a dramatic new moment.

Narrative twist

Example:

Change the storyline of @Video1 so the dramatic scene becomes comedic.

Style transfer

Input: video + style reference

Example:

Recreate @Video1 using a black-and-white ink painting style.

Commercial recreation

Input: popular advertisement + your product image

Example:

Follow the concept and rhythm of ​**@Video1​’s advertisement, but replace the product with ​@Image1**​.

Multi-character replacement

Input: fight video + two character images

Example:

Replace the two fighters in @Video1 with @Image1 and ​**@Image2**​.

Music video from still images

Input: multiple photos + music video reference

Example:

Synchronize @Image1 to @Image6 with the rhythm and beat from ​**@Video1**​.

Pro Tips

1. Use High-Quality Reference Materials

If your reference @Image1 is blurry, the generated video will likely also appear low quality. For best results, use 2K or 4K high-resolution images. Clean source materials help Seedance 2.0 better understand visual details.

2. Be Explicit About What You Want to Reference

Instead of just mentioning a file, clearly describe what aspect should be extracted.

For example:

• Camera movement
• Choreography
• Rhythm
• Visual effects
• Transitions
• Voice tone or timbre

Example:

Reference @Video1’s camera movement and transition style.

This provides much clearer guidance than simply naming the file.

3. Combine Video and Image References

You can mix different media types to achieve motion transfer.

Example:

@Image1 performs the dance from @Video1.

This allows you to create a custom character while preserving motion choreography.

4. Iterate with Small Adjustments

You don’t need to rewrite the entire prompt if the result is not ideal.

Try:

• Changing one word
• Swapping one reference asset
• Adjusting motion speed or camera angle

Seedance 2.0 is fast enough to support rapid experimentation.

5. Clearly Distinguish Edit vs Reference Tasks

Make sure to specify whether you want to:

• Modify an existing video
• Use the video only as creative reference

These are handled differently by the model.

6. Label Assets Carefully

When working with multiple files, double-check that each reference tag is correct.

Mixing up @Image or @Video numbers is one of the most common causes of unexpected results.

7. Use Time Segments for Longer Videos

For videos around 10–15 seconds, organize prompts by time segments.

Example:

0–3s: Close-up of character face. 4–8s: Camera pulls back to reveal environment. 9–12s: Action sequence with tracking shot. 13–15s: Closing frame with brand text.

This gives you more precise creative control.

8. Reference Audio from Videos

You do not need to upload separate audio files.

Example:

Background music references @Video1’s audio.

Seedance 2.0 can extract background music, environmental sound, and rhythm cues directly from video.

9. Quick Recap Best Practices

• Write actions using slow, continuous motion descriptions
• Keep camera movement simple and stable
• Add visual stability constraints
• Add quality and style keywords at the end of prompts
• Avoid overly complex multi-person interactions
• Clearly label reference roles

What to Avoid

Avoid Complex Multi-Person Actions

Such as:

• Large fight scenes with many characters
• Extremely complex choreography involving multiple interactions

These can reduce motion consistency.

Avoid Vague Descriptions

Phrases like:

• “Looks good”
• “Very beautiful”
• “Super cool”

These do not provide enough guidance for generation.

Avoid Contradictory Requirements

Example:

• “Ultra-fast motion” + “extremely stable footage”

Choose one primary direction.

Avoid Reference Confusion

Do not mix up:

• Image references
• Video references
• Character identity references

Avoid Restricted Content

Do not upload or generate:

• Explicit or violent scenes
• Dangerous activities
• Copyrighted character IPs
• Restricted real human facial data where not permitted

(Some guide examples may use controlled or licensed reference materials.)

Ready-to-Use Prompt Templates

Template 1 — Cinematic Portrait Video

A young woman walking slowly through a forest, gentle wind moving her hair, soft smile, warm sunlight, medium shot, slow push-in camera movement, smooth footage, 4K HD, cinematic lighting, stable motion, natural body proportions, clear facial details.

Template 2 — Atmospheric Landscape

Seaside sunset, waves gently moving toward the shore, slow horizontal camera pan, warm orange lighting, calm and peaceful atmosphere, silky smooth footage, 4K ultra HD, no flickering, no ghosting.

Template 3 — Image to Video (First Frame Lock)

Use the reference image as the first frame. Maintain consistent character appearance and clothing. Character slowly raises hand and turns around. Natural, fluid motion, no stiffness, stable camera movement, cinematic quality.

Template 4 — Product Commercial

Reference @Video1’s camera work and rhythm. Replace the product with @Image1. Smooth tracking shot with close-up product details. Cinematic lighting, 4K ultra HD, professional commercial quality.

Template 5 — Character Dance Transfer

@Image1 performs the dance from @Video1. Maintain face consistency and character identity. Fluid choreography, medium shot, slight orbit camera movement, cinematic lighting.

Template 6 — Narrative Scene With Dialogue

The man from @Image1 walks through a corridor. He takes a deep breath at the door, unlocks it, and enters. His daughter and pet dog run toward him and hug him. Warm indoor lighting. Natural dialogue with ambient home sound.

Template 7 — One-Take Shot

@Image1 as the first frame. Continuous one-take shot. Camera moves through doorway into room @Image2, pans to corridor @Image3, follows character to window @Image4, then pushes outside to @Image5. No cuts, smooth motion

Template 8 — Video Extension Ending

Extend @Video1 by 10 seconds. Camera slowly pushes in while scene fades to clean background. Product appears with spotlight lighting. Brand text fades in: “Your Brand — Tagline”.

Template 9 — Beat-Synced Montage

@Image1 through @Image6 synced to @Video1 rhythm. Images change on musical beats. Strong visual tension, dreamlike style, dynamic framing, dramatic lighting changes.

Template 10 — Emotional Performance Scene

@Image1 as the first frame. Character slowly looks up. Emotion shifts from calm to intense. Camera rotates and pushes in during emotional peak. Reference @Video1 for emotional expression style.

Settings Reference

Final Thoughts

Seedance 2.0 is designed to give creators real creative control, not just automated generation.

Instead of simply typing prompts, you can direct the model by specifying:

• Which asset controls motion
• Which asset controls style
• Which asset controls audio
• Which asset controls character identity

The more precise your direction, the more unique your output will be.

The multi-modal system allows you to freely combine:

• Motion from dance videos
• Camera language from films
• Voice tone from audio sources
• Visual style from paintings or photos

Use natural language to describe your vision.

Remember: the best content is not just what the AI can generate — but what only you can imagine.