Alibaba Cloud · 最新統一型號
Generate 2–30 second video with synchronized audio from text, images, first/last frames, or combined image + video + audio references — and turn documents and web pages directly into video. Up to 1080p with adaptive aspect ratios.
精確模型工作區
The generator is pinned to wan3.0-video. Available inputs, controls, and charged credits come from the live route.
型號概覽
Wan 3.0 is Alibaba's all-modality reference video model. Beyond the text, image, and endpoint workflows of earlier Wan versions, it accepts documents and web pages as creative sources and combines image, video, and audio references in a single request — useful when a clip must match existing brand assets, a voice track, or written source material.
廣域網路模型能力
Feed a document or web page as reference and Wan 3.0 turns its content into a coherent explainer clip — a workflow unique among commercial video models for course, tutorial, and product-documentation video.
廣域網路模型能力
Attach up to 10 reference images, 5 reference videos, and 5 audio clips together. Each asset guides identity, motion, or soundscape so the output stays consistent with your existing material instead of reinventing it.
廣域網路模型能力
Define the opening and closing composition with two frames, then let Wan 3.0 direct the motion between them. The 2–30 second range covers everything from micro-loops to full scene beats without stitching.
基於任務的範例
從您需要的可交付成果中選擇一個工作流程，然後將隨附的提示啟動器調整為一個清晰的鏡頭。
Turn lecture notes, documentation, or web articles into narrated video segments.
Turn the attached document into a clear explainer video. Open on [hook visual], walk through [key points], close on [summary frame]. Calm professional narration.
Combine product images, a motion reference, and brand audio into one coherent clip.
Use the attached product images for exact appearance and the audio for pacing. The product [action] in [setting]; end on the hero composition from reference image 1.
Generate 2–5 second seamless loops or vertical hooks at 9:16 with synced sound.
Seamless 9:16 social loop of [subject] [action]. Rhythmic motion synced to the audio reference, clean silhouette, loopable start and end frames.
AI視訊模型對比
將此用作模型選擇指南，而不是替代即時控制。公共規範和連接的提供者路由可以公開不同的參數子集。
| 特點 | Wan 3.0 | Seedance 2.0 Pro | Veo 3.1 | Kling 3.0 |
|---|---|---|---|---|
| 最適合 | Document & webpage-to-video explainers · All-modality reference consistency · 2–30s clips up to 1080p with audio | 多模式講故事和連續性主導的創作 | 電影寫實主義和精美的視聽場景 | 角色動作、動作與創作者工作流程 |
| 連接輸入 | Generate 2–30 second video with synchronized audio from text, images, first/last frames, or combined image + video + audio references — and turn documents and web pages directly into video. Up to 1080p with adaptive aspect ratios. | 文字·圖像·多模態參考 | 文字·圖片 | 文字·圖像·參考工作流程 |
| 場景控制 | Documents and web pages become video sources · All-modality references in one request · First / last frame plus 2–30 second range | 多鏡頭和多模式方向 | 鏡頭解釋和視聽指導 | 動作控制和角色表演 |
| 音訊工作流程 | Synchronized | 路線報告音訊功能；確認即時控制 | 原生視聽工作流程 | 音訊和口型同步工作流程因路線而異 |
| 選擇何時 | Course & explainer video · Brand-consistent product film · Social loops & hooks | A multimodal story workflow and continuity-led creation | 最終的電影視聽打磨最重要 | 角色表現與動作控制引領劇情簡介 |
PixMind 生產 API 快照已驗證 2026-07-13。現場發電機控制是最終的。
快速啟動
命名主題、一個動作、攝影機路徑、光線、聲音和結局。僅新增所選輸入模式所需的參考職責。
Turn the attached document into a clear explainer video. Open on [hook visual], walk through [key points] with [visual metaphor], close on [summary frame]. Calm professional narration, clean motion graphics aesthetic.
使用此提示Use the attached product images for exact appearance, the motion reference for camera rhythm, and the audio for pacing. The product [action] in [setting]; lighting matches [mood]; end on the hero composition from reference image 1.
使用此提示Move naturally from the supplied first frame to the last frame. The subject [action] while the camera [movement]; keep identity, wardrobe, and environment consistent; finish exactly on the end-frame composition.
使用此提示使用文字進行創作，使用影像進行身份或構圖，使用第一幀和最後一幀作為受控端點，以及使用運動、攝影機節奏或聲音的參考。
某些連接的路線報告音訊功能，但具體控制因型號模式而異。在發布之前確認實時生成器中選定的路線並檢查對話、時間表和工件。
商業用途取決於提供者條款以及您對每個提示、參考、身分、聲音、標誌和品牌資產的權利。在發布之前查看輸出和當前條款。