Seedance 2.0 Mini turns a shot brief, storyboard frames or media references into a silent 4–15 second video at 480p or 720p. Develop the visual sequence first, then add dialogue, music and sound in your editing workflow.
set up https://beatapi.io/SKILL.mdGive this to your agent
Not available on free credit — top up to unlock it.
Generated result
Run the playground to see output here.
Current BeatAPI public retail rates for this model.
| Model | Specifications and pricing | Price |
|---|---|---|
| Seedance 2.0 Mini | 8 sec · 480p · no video input | $0.279931 |
| Seedance 2.0 Mini | 8 sec · 480p · with video input | $0.339902 |
| Seedance 2.0 Mini | 8 sec · 720p · no video input | $0.602001 |
| Seedance 2.0 Mini | 8 sec · 720p · with video input | $0.730973 |
Public BeatAPI video model contract for prompt, frame, and reference-driven generation.
seedance-2.5Public BeatAPI video model contract for prompt, frame, and reference-driven generation.
seedance-2Faster official Seedance 2.0 route for 480p and 720p production drafts.
seedance-2-fastLower-cost official Seedance 2.0 route for high-volume 480p and 720p drafts.
seedance-2-mini| Shot requirement | Seedance 2 Mini | Seedance 2 Fast | Seedance 2 Standard |
|---|---|---|---|
| API | Seedance 2 Mini | Seedance 2 Fast | Seedance 2 Standard |
| A product reveal with an existing voiceover | Mini: use a product first frame and one camera-move brief; 4–15s at 480p/720p, then lay the existing sound over the silent result. | Fast: keep the same frame workflow when the draft also needs generated audio. | Standard: keep the visual anchors when the delivery needs 1080p/4K rather than a 720p draft. |
| A transition between two storyboard panels | Mini: two images define the first and last frames; describe the action connecting them. Do not mix this mode with reference arrays. | Fast: the same two-frame mode plus optional audio; no claim of a faster completion time. | Standard: the same start/end framing with higher-resolution options; resolution alone does not guarantee a better transition. |
| A motion reference for a vertical shot | Mini: choose reference mode for image/video guidance; videos total at most 15s. Audio references need visual references and do not create output audio. | Fast: use the separate reference mode for an audiovisual draft; include reference-video length in the estimate. | Standard: reference mode serves a higher-resolution shot; compare the full reference workload and its billing floor, not just output seconds. |
| Move from a draft to delivery | Mini: choose when the edit supplies sound and 480p/720p is sufficient. The 5s serum example is silent and demonstrates only this visual task. | Fast: choose for 480p/720p scenes that need generated sound; the shared Mini example is not evidence for this. | Standard: choose when the brief explicitly requires 1080p/4K; retain task records and assess the finished shot in the edit. |
| Text-only shot | Send prompt without any media arrays. Describe the visible subject, action, environment and camera; audio instructions do not enable a soundtrack. |
|---|---|
| First and last frames | Use images with one first-frame URL or two first/last-frame URLs. Keep this mode separate from all reference_* arrays. |
| Multimodal visual draft | Use up to 9 reference images, 3 videos or 3 audio files. Audio requires an image or video reference. Reference videos have a combined 15-second limit and contribute to billing; the output remains without generated audio. |
Seedance 2 Mini API makes short visual scenes from a written brief, storyboard frames or reference media. Its three input routes serve different creative decisions: invent a shot, connect planned compositions, or carry visual direction into a new take. Mini outputs 4–15 seconds at 480p or 720p without generated sound.
A subject, an action and a camera brief can become a moving shot without a supplied image. A toy character walking through a city or a staged sports moment can begin as text. A short, focused event fits the 4–15 second envelope; multiple competing actions, precise lettering and complicated anatomy still need review in the resulting clip.
One image can establish the opening frame, while two images establish the opening and ending. A packshot can become a reveal, or two storyboard panels can become a transition. Specify what moves between those anchors. Frame input is a separate mode from reference arrays, and an anchor does not guarantee that every product detail or face remains unchanged.
Reference images and videos can guide subject appearance, scene direction or motion. Use reference mode when a visual example matters more than matching two endpoint frames. Video references add billable duration and have a combined 15-second limit. Audio references require an image or video reference; they do not enable a generated soundtrack on Mini.
A woman holds a small white serum bottle up toward her phone camera and turns it slowly so the label faces the lens. Soft daylight from a window, a bare bedroom wall behind her. Vertical selfie framing, handheld, natural skin texture.
Cinematic shots. Different camera angles. A cool-dressed rapper toy sitting on the cap of a man. The man walks through the city as the living toy on his cap is vibing to a whistling song.
A beautiful, photorealistic young adult English woman with long flowing brown hair, wearing a red-and-black fantasy warrior outfit, runs through an ancient misty forest. Suddenly, multiple enormous fire-breathing dragons emerge behind her and attack. She instantly draws her gleaming steel sword and fights fearlessly at incredible speed. She performs breathtaking high jumps, aerial spins, lightning-fast sword strikes, agile flips, and powerful midair attacks. She dodges streams of dragon fire by leaping sideways and counterattacks with precise, fluid sword movements. One by one, the dragons are defeated in spectacular cinematic battles. The final dragon lunges at her, and she jumps high into the air, delivers a decisive finishing strike, and lands gracefully as the dragon collapses behind her. Her hair and clothing move naturally with every action. Ultra-realistic human anatomy, authentic facial expressions, believable physics, seamless fast-paced choreography, detailed dragon scales, cinematic tracking shots, dramatic lighting, flying sparks, smoke, fire, and realistic sound effects. Original fantasy scene, live-action movie realism, 4K quality, no cartoon style, no artificial-looking characters, no blood or gore. Vertical 9:16, 10 seconds, continuous action, extremely fast combat, consistent character appearance.
主体： 红方拳击手，20岁东亚女性，短的金色头发，穿着红色运动背心，红色拳击手套，红色拳击短裤。 蓝方拳击手，20岁东亚女性，黑色长直发，穿着蓝色运动背心，蓝色拳击手大，蓝色拳击短裤。 场景： 一个拳击俱乐部的比赛拳台，手持跟随拍摄，明显的抖动 风格： 运动摄影，中景，浅景深，慢镜头，展示红方拳击手击倒蓝方拳击手的最后时刻 运动： 1、荷兰角构图，红方拳手蓄力，一记重重的直拳打向右下角的蓝方拳手，蓝方被击中面部，蓝方躲着退到护栏边靠在护栏上，举起双拳防御。 2、红方拳手前进一步，快速连续击打四下，蓝方用手拳努力挡下，都挡住了，但重心明显不移，身体左移想离开击打范围，镜头晃动跟随拍摄，展示现场感。 3、红方拳手快速小碎步调整角度虚晃一招，蓝方中计，用力一个左手摆拳想击中红方，红方机灵的一闪避开了。 4、蓝方左手方位后侧已完全暴露在红方视线范围内，红方一记重重的右摆拳，慢镜头，击中蓝方的左耳，蓝方痛苦地避让。 5、继续慢镜头红方接着一个左手勾拳，击中蓝方的下巴，蓝方脸被打变形，立即恢复正常速度拍摄，蓝方向后倒去重重摔在台边，镜头晃动，裁判连忙过来，阻挡了红方的继续进攻 现场观众声音随着比赛进展而呐喊，非常热闹，拳击擂台赛场，专业拳击服与护具，动态肌肉细节，汗水飞溅，真实光影，电影质感，高细节，流畅动作，极致慢动作，张力十足，紧张氛围，写实风格，8K，高清，动态景深 负面：模糊，卡顿，扭曲肢体，变形人体，多余人物，水印，文字，卡通画风，低画质，违和光影 无文字，无字幕，无音乐，保留拳击比赛现场音效，16:9
Create the moving images your edit still needs: a product opening, a scene transition, a playful character moment or an action beat. Mini supplies individual visual clips; your editing workflow assembles the story, titles and sound.
A product photograph and a slow turn, push-in or reveal brief become a short shot for an advert that already has narration. The moving image introduces the item while the existing edit carries the spoken offer. Packaging, logos and label readability remain part of the final visual check.
Two planned frames and the movement between them become an animatic segment for a pitch or shot review. A team can discuss when the reveal lands and how the next scene opens before committing to production. The clip is a planning deliverable, not a promise that the final shoot will match every generated detail.
A compact idea such as a tiny living toy riding on a person’s cap becomes a short character-led scene. Camera angles and the character’s visible action give the post a clear visual hook. Mini creates the clip; captions and any music are added in the social edit.
A boxing exchange or fantasy encounter becomes a visual candidate for discussing movement, camera position and the turning point of a scene. Limit the brief to the action that must read clearly. Fast contact, limbs, props and continuity can need another take; the clip does not replace choreography or safety planning for a real shoot.
An owned illustration or pet photo can anchor a short movement or visual joke for a personal post. The supplied image defines the appearance you want, while the brief describes the event. Keep the reference file with the result and avoid treating a changed face, fur pattern or scale as an exact reproduction.
Keep the product and intended message fixed while varying the opening movement or composition. The resulting clips give an editor candidates for a vertical post or campaign cut. Comparing them is a creative selection exercise; conversion gains and success rates require your own campaign evidence.
| Shot requirement | Seedance 2 Mini | Grok Imagine Video 1.5 |
|---|---|---|
| API | Seedance 2 Mini | Grok Imagine Video 1.5 |
| First and last compositions must both be anchored | Mini supports two start/end frames; use it for a silent storyboard transition. | Grok 1.5 accepts one first frame, not the Mini two-frame mode; choose it when only the opening needs an image anchor. |
| The brief depends on a motion-reference video | Mini accepts video references in its separate multimodal mode, with a 15s combined cap and added billable input. | Grok 1.5 exposes image references, not reference-video/audio arrays; use a text motion brief rather than transplanting Mini fields. |
| The edit needs a very short cutaway or 1080p shot | Mini stays within 4–15s and 480p/720p, with sound supplied by editing. | Grok 1.5 accepts 1–15s and up to 1080p; at 1080p at most one image is allowed. Its generated sound is a separate workflow condition. |
| Compare one deliverable, not the lowest tier | For an 8s 720p text-only estimate, use the Mini no-video price row; adding a video reference changes the workload. | Use an 8s 720p no-reference row for the same output length. Inputs, audio and final settlement may differ, so cost alone is not a quality verdict. |
Create a silent video task with one input mode. Save its ID, retrieve the finished media and keep its references and usage alongside the draft.
Sign up or sign in to open the Dashboard. Check your balance and open the credits page if you need to add funds.
Open DashboardMini fits an edit that needs more moving-image candidates while its sound and final assembly remain separate.
Silent clips suit an existing voiceover, music bed or animatic review. There is no need to make a soundtrack part of this generation request when the edit already provides it.
Text creates a scene from scratch, start/end frames connect approved compositions, and reference mode guides a new take. The choice follows the brief rather than a claim that one input mode always produces better work.
Mini has separate resolution and video-reference price conditions. A short output with a long reference is not the same workload as text-only output. Existing task records let your application retain the prompt, inputs and settled usage for each candidate.
Seedance 2.0 Mini is the Mini video model in ByteDance’s Seedance 2 family. BeatAPI accepts a text brief, one or two frame images, or multimodal references and returns a 4–15 second visual clip at 480p or 720p through an asynchronous task. Mini does not generate an audio track. It is useful for individual product shots, animatic segments and creative candidates that will enter an edit; it does not automatically produce a finished campaign with captions, licensed music and a complete story.
Connect Seedance 2.0 Mini with other production-ready APIs in the Beat API catalog.
Guides and practical notes to help you get more from Seedance 2.0 Mini.
