ByteDance video model on Lovart ME
Seedance 2.0 Mini AI Video Generator
Direct a complete 4-15 second clip with text, frames, visual references, and sound cues.
Seedance 2.0 Mini uses a unified multimodal workflow: start from a written scene, anchor the opening and ending with images, borrow motion from a short video, or guide the soundtrack with audio. Choose the input that carries useful production information instead of adding references without a clear role.
- 4-15s
- Flexible duration
- Select any whole-second duration in the supported range.
- 720p
- Maximum output
- Use 480p for iteration or 720p for the final clip.
- 7
- Aspect ratios
- Landscape, portrait, square, ultrawide, and adaptive options.
- A/V
- Native audio
- Generate synchronized dialogue, ambience, and sound effects.
01 / MODEL
What Seedance 2.0 Mini is built to control
The useful difference is not simply accepting more files. Each input can carry a separate part of the direction, which makes a short clip easier to plan and revise.
Multimodal scene direction
Combine a prompt with images, video, and audio references. Name each reference in the prompt and state whether it controls subject identity, composition, motion, camera language, rhythm, or sound.
Stable motion and interaction
Describe the action as a sequence with clear subjects, direction, speed, and contact points. This gives the model stronger constraints for body movement, object interaction, and camera motion.
Short multi-shot storytelling
Plan several connected shots inside a clip of up to 15 seconds. Mark each shot, its framing, and the transition so the sequence has a readable beginning, development, and finish.
Synchronized audio-video output
Ask for dialogue, room tone, music cues, and sound effects in the same brief. Audio can be disabled when the result will enter a separate sound-design workflow.
02 / INPUTS
Give every reference one clear job
More references do not automatically improve a result. Use the smallest set that resolves a real uncertainty in the scene.
Text prompt
Define the deliverable, subject, action, setting, camera, light, pacing, and required audio. Use numbered shots when the scene changes during the clip.
First and last frames
Use the first image to anchor the opening composition and the second to define the destination frame. Describe the transition between them rather than only describing both still images.
Reference images
Use extra images for a character, product, environment, palette, or visual treatment. Keep their roles distinct and refer to them as @Image 1, @Image 2, and so on.
Reference video and audio
A short video can guide motion, camera movement, or timing; an audio clip can guide voice, rhythm, ambience, or effects. Each media group supports up to three files totaling no more than 15 seconds.
03 / OUTPUT
Choose output settings around the delivery channel
Set duration, resolution, and ratio before refining the prompt. These choices affect composition, pacing, and the credit total shown in the generator.
480p for direction tests
Use 480p to compare motion, shot order, timing, and reference choices at a lower credit cost. Move to 720p after the scene structure works.
720p for approved clips
Choose 720p when the framing and motion are ready for closer review or publishing. Inspect faces, hands, edges, brand marks, and small text before use.
Ratios from 21:9 to 9:16
Use 16:9 or 21:9 for wide scenes, 1:1 for balanced social placements, and 9:16 for full-screen mobile. Adaptive lets the model follow the strongest visual input.
04 / PROMPTS
Seedance 2.0 Mini prompt templates for production tasks
Replace the bracketed fields and remove instructions that do not affect the deliverable. Keep reference labels stable from one revision to the next.
Product motion spot
Use a product image to preserve shape and a short motion reference to define the camera move.
Create a [duration]-second [ratio] product spot. Keep the product shape, label, and colors from @Image 1. Follow the smooth orbit and acceleration pattern in @Video 1 without copying its setting. Scene: [surface and background]. Light: [direction and quality]. End on a clean hero frame with space on the [side]. Audio: [ambience and one product sound].Production note: Keep logos and label text large enough to inspect in the result.
Dialogue scene
Plan a compact exchange with explicit speaker, framing, and room tone.
Create a [duration]-second two-shot conversation in [location]. Shot 1: [character A] says '[line]' in a calm voice, medium close-up. Shot 2: [character B] reacts, then says '[line]', over-the-shoulder framing. Preserve appearance from @Image 1 and @Image 2. Natural pauses, consistent eyelines, subtle [room ambience], no subtitles.Production note: Use short lines and verify lip sync and exact wording before publishing.
Vertical social sequence
Build a readable three-shot story for a 9:16 feed without overcrowding the frame.
Create a [duration]-second 9:16 social video about [topic]. Shot 1, 0-3s: immediate visual hook, [action]. Shot 2, 3-[time]s: close detail showing [benefit]. Shot 3: clear final reveal with clean space in the upper third. Fast but legible cuts, handheld energy from @Video 1, bright natural sound, no generated text.Production note: Leave on-screen copy for editing software when spelling must be exact.
Frame-to-frame transition
Direct how the first composition should evolve into the final one.
Animate from @Image 1 to @Image 2 over [duration] seconds. The camera [movement] while [subject action] causes the transition. Preserve the subject identity, clothing, and color palette. Midpoint: [specific intermediate state]. Finish on the exact composition of @Image 2 and hold for one second. Audio: [transition sound and ambience].Production note: Choose frames with a plausible visual path between them.
05 / WORKFLOW
How to use Seedance 2.0 Mini on Lovart ME
Treat each generation as a small production brief. The same four steps work for text-only clips and reference-led scenes.
- 01
Choose text or image mode
Start with Text to video for a scene built from the brief. Choose Image to video when an opening frame or visual identity must anchor the result.
- 02
Add only useful references
Upload frames, extra images, a motion clip, or an audio cue. Keep video and audio references within three files and 15 total seconds for each type.
- 03
Set duration, format, and audio
Choose 4-15 seconds, 480p or 720p, and the destination aspect ratio. Leave synchronized audio on when sound is part of the brief.
- 04
Generate and review the whole clip
Check shot continuity, motion, identity, hands, lip sync, small text, and sound timing. Revise one constraint at a time so the effect of each change is clear.
06 / FAQ
Seedance 2.0 Mini questions
These answers reflect ByteDance's published Seedance 2.0 capabilities and the controls available in the current Lovart ME integration.
What is Seedance 2.0 Mini?
Seedance 2.0 Mini is a ByteDance video-generation model exposed through Lovart ME via Kie.ai. It accepts text and several reference types, produces 4-15 second clips, and can generate synchronized audio.
Can I create a video from an image?
Yes. Choose Image to video and upload at least one image. The first image anchors the opening frame, the second can guide the ending frame, and additional images can provide visual references.
How many reference videos and audio files can I use?
You can add up to three reference videos and three reference audio files. The videos must total no more than 15 seconds, and the audio files must separately total no more than 15 seconds.
Does Seedance 2.0 Mini generate sound?
Yes. Synchronized audio is enabled by default in this workflow. You can turn it off when you plan to create dialogue, music, ambience, or effects in a separate audio process.
Which resolutions and aspect ratios are supported?
The current integration supports 480p and 720p output. Ratios are 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, and adaptive.
How are credits calculated when I upload a video?
Without a reference video, the selected resolution rate is multiplied by output duration. With a reference video, a different rate is multiplied by the combined reference-video duration and output duration. The live control shows the rounded total before generation.
Can I use generated videos commercially?
Usage depends on Lovart ME terms, the rights attached to your uploaded references, and applicable law. Review the current terms and clear all source material before commercial distribution.
07 / NEXT
Continue with the right video workflow
Compare the lighter Seedance option, return to the full model selector, or study public examples before the next generation.