將照片和插畫轉成視頻
從主體清晰的人像、產品圖、插畫或生成畫面開始。
AI 照片動畫
用 AI 讓一張照片動起來
上傳一張照片，描述主體和鏡頭如何運動，即可將人像、產品圖或插畫變成短視頻。
AI 圖片轉視頻功能
AI 圖片轉視頻會把上傳圖片作為起始影格和視覺錨點，再產生後續畫面。
用提示詞控制主體動作與運鏡
把主體動作和鏡頭行為寫成兩組獨立指令。
選擇圖片轉視頻模型
比較目前可用且支援圖片輸入的視頻模型。
圖片轉視頻案例與提示詞
對照參考圖片、完整提示詞和已發布的視頻結果，挑選一個方向，再依自己的圖片改寫。
高級香水廣告視頻續寫提示詞
Synthia視頻提示詞（圖片 → 視頻延續）： 一位擁有上傳圖片面容的年輕女子，身穿俐落的黑色晚裝，站在燈光柔和的奢華精品店內，背景映照着金色光澤與鏡面。她輕輕將一瓶高級香水舉至頸旁，按下噴頭，讓細緻的香水薄霧飄散於空氣之中。在溫暖的電影感燈光下，細膩的香氛水霧閃爍生輝。她短暫閉上雙眼，細意享受香氣，然後對着鏡頭溫柔微笑，並說道：「I just love it.」優雅時尚廣告風格、動作流暢自然、淺景深、電影級燈光、超寫實、奢華美妝廣告氛圍。查看完整提示詞僅替換提示詞，保留你的圖片和設置。請依自己的圖片調整內容。
15 Second Coffee CM Japan to Italy Trip
BiancaUltra-realistic live-action commercial, 15 seconds, 16:9, 480p, cinematic, high quality, natural colors, movie-like lighting. ■Overall Tone: Stylish, bright, and lively world like an Italian bar. BGM is light and sophisticated Italian-style espresso jazz (accordion + piano + light percussion). Emphasize morning freshness and happiness. ■Characters: Japanese woman, 25 years old, living alone, natural and cute, slightly rough hair from waking up, light makeup, loungewear (simple and clean clothes). ■Reference: Canned coffee image Image 1 (completely reproduce the design and texture of this can). ■Scene Composition: ① (0-4s) Morning in a Japanese 1K apartment. Small kitchen with soft morning sunlight. Canned coffee (same as reference image) on a table with a slight lived-in feel. Woman wakes up groggily and picks up the canned coffee in a natural movement. Camera slowly zooms in (dolly-in), cinematic. ② (4-7s) Beautifully emphasize the moment the woman pops open the canned coffee in slow motion. Not carbonated, but an effect like aromatic steam at the moment of opening. Sound: Pleasantly emphasize the opening sound. ③ (7-13s) ★Main Scene (Long & Glamorous): The moment it's opened, space changes seamlessly. The Japanese kitchen transforms into a glamorous Italian bar while being enveloped in light. - Cobblestones - Colorful buildings - Espresso machine - Showcase with bread and sweets - Lively people (smiles, conversation) - Warm and bright sunlight. Camera follows the spatial change in a smooth one-cut (transition is natural like magic). Woman is surprised → immediately changes to a happy expression. Woman takes a sip of the canned coffee → A truly delicious-looking, natural, and cute smile → Narrowing eyes slightly in a happy reaction. Combine slow motion and normal speed to emphasize the sense of happiness. Light is golden, soft, and beautiful. ④ (13-15s) Canned coffee floats in the air and rotates slowly (high-end product shot). Controlled so the front label of the can always faces the camera. Beautiful reflections, lighting, and highlights. Background is a slightly blurred Italian cityscape. ■Narration (Young woman's voice, bright and light): 'Once you open it, you're in Italy♪ Via Espresso.' ■Additional Specifications: - Prioritize live-action quality (exclude CG feel). - Skin, light, shadows, and reflections are all real. - Camera work is smooth and cinematic. - Use depth of field appropriately. - Colors are warm to evoke appetite and happiness. - Overall visual expression that 'conveys the scent.'查看完整提示詞僅替換提示詞，保留你的圖片和設置。請依自己的圖片調整內容。
Voronoi Lamp Product Animation
ShamsExecute the uploaded image, using every frame carefully keep the product intact 100%.查看完整提示詞僅替換提示詞，保留你的圖片和設置。請依自己的圖片調整內容。
Anime Witch Transformation with Bat Wings
Mistress Belladonna Nightshade"Image1 is the reference for the entire video. It is a sketch art anime style. 1sec-3seconds the woman is standing still and the wind is blowing hard. Her dress and hair are blowing in the wind. 4sec-6sec the woman magically causes the staff in her hand to disappear. She smiles with an evil smile. 7sec-15sec She holds out her arms and large bat wings appear from her back draping down from her arms. (the wings should be attached to her as the wings are attached to a bat). She turns her back to the camera and jumps off the cliff flying towards the castle in the distance, wings flapping like a bat. She should have bat wimgs and not bird wings. The sketch art anime style should be used throughout the video just like the reference image1"查看完整提示詞僅替換提示詞，保留你的圖片和設置。請依自己的圖片調整內容。
Flying Witch Departure
akiraA Japanese girl around 13 years old. Short black hair with a delicate adolescent expression, mixing anxiety and expectation. Wearing a simple black A-line knee-length dress with slightly puffed long sleeves and no decoration. Black tights and simple flat shoes. A large red ribbon on her head. The girl is straddling a bamboo broom, sitting firmly with both legs. She is leaning forward, gripping the handle with both hands, clearly maneuvering and flying the broom. The setting is a quiet residential area at night. A European-style house on a green hill with white walls, wooden textures, a triangular roof, and a chimney. The house sits normally on the ground surrounded by gardens and greenery. A full moon in the sky illuminates everything with soft moonlight. The girl slowly floats up from the eaves of the house and ascends into the night sky. The ribbon and skirt flutter naturally in the wind. The camera follows smoothly from behind, capturing the movement from floating to ascending in a single sequence. Realistic yet cinematic, nostalgic, and slightly fantastical tone.查看完整提示詞僅替換提示詞，保留你的圖片和設置。請依自己的圖片調整內容。
Pixar Style Robot Adventure
FANTAN GLITCHBOX[STYLE] Heartwarming cinematic 3D animation, emotional sci-fi adventure, soft lighting, detailed rendering, warm family atmosphere, animated movie quality [CHARACTERS] Small white child robot with glowing blue eyes Large white parent robot, gentle and protective appearance [SCENE] A warm futuristic world filled with nature and ruins. Soft sunlight, emotional storytelling, beautiful landscapes, cinematic camera movement. [SEQUENCE] The parent robot gently prepares the small robot for a journey inside a cozy mechanical home. The child robot steps outside into a bright unknown world for the first time. Wide cinematic shots: mountains, ancient ruins, grass moving in the wind, beautiful skies. The small robot climbs cliffs, crosses rivers, and explores mysterious environments. The robot meets a tiny forest creature and carefully helps it. Night falls. A storm begins. Heavy rain and lightning surround the small robot. The child robot looks frightened and alone. The robot remembers the warmth of the parent robot. The small robot stands up again and continues forward with determination. The child robot discovers a glowing crystal or ancient object in beautiful ruins at sunrise. After the adventure, the robot returns home. The parent robot kneels and opens its arms. The small robot runs forward and is tightly embraced. Final emotional close-up: the parent robot hugging the child robot gently, warm sunset light, soft glowing eyes, peaceful ending. [CAMERA] cinematic tracking shots, slow emotional camera movement, wide landscape shots, soft close_ups, gentle depth of field [EFFECTS] warm sunlight, soft particles, rain, wind, subtle lens flare, cinematic atmosphere [MOOD] heartwarming, emotional, hopeful, family love, gentle adventure查看完整提示詞僅替換提示詞，保留你的圖片和設置。請依自己的圖片調整內容。
角色圖片生成動作視頻
用一張角色圖製作節奏更快的動作鏡頭，讓身體、背景和鏡頭共同產生運動，同時保持臉部與服裝可辨識。
人像照片生成動態視頻
為靜態人像加入輕微表情、頭髮或布料運動、粒子效果和緩慢鏡頭移動。
靜態插畫生成動畫視頻
讓插畫或 AI 作品動起來，同時不重建原有構圖。
使用方法
如何用 AI 將圖片轉成視頻
有效的圖片轉視頻流程會從乾淨的起始畫面開始，分層描述運動，並在增加複雜動作前先評估第一版結果。
01
上傳照片或圖片
選擇一張主體、裁切和風格都符合預期的圖片。
上傳照片或圖片
使用清楚的照片、插畫、產品圖或生成作品。02
撰寫動態提示詞並選擇視頻模型
先描述主體最重要的動作，再補充鏡頭運動、速度和氛圍。
描述你想要的視頻效果
選填讓畫面中的人物自然眨眼並微微轉頭，頭髮隨風擺動，鏡頭緩慢推進
長寬比
質素
自動
高
中
低
參考圖模式
自動
首尾幀
參考圖
聲音
自動
有聲
無聲
03
生成 AI 視頻
開啟準備好的圖片轉視頻任務並產生第一版。
圖片轉視頻常見問題
哪些圖片更適合生成 AI 視頻
建議使用主體清楚且與背景有明顯區分的圖片。照片、插畫、產品圖和生成作品都能獲得不錯的效果。
如何用 AI 讓一張照片動起來
上傳一張照片，選擇運動參考，或分別描述主體與鏡頭應該怎麼動，再選擇模型。如果手上沒有圖片，想直接從場景描述開始，可以使用AI 文字轉視頻生成器。
產品圖和插畫可以製作成 AI 視頻嗎
可以。產品圖、海報、插畫和 AI 作品都能作為起始畫面。產品鏡頭可以嘗試簡單的推進、環繞或輕微背景運動；插畫更適合保持構圖穩定，只讓光線、煙霧、水面等指定元素變化，從而減少不必要的畫面重建。
圖片轉視頻一定要寫提示詞嗎
不需要，只上傳圖片也能開始。補充主體動作、鏡頭方式、節奏或氛圍等簡短描述，可以給模型更明確的方向。
怎樣寫出更有效的圖片轉視頻提示詞
可以按照主體、主要動作、鏡頭運動、速度和氛圍的順序來寫。先從一個清晰動作開始，例如人物緩慢轉頭，鏡頭輕輕推進，微風帶動頭髮。如果結果不穩定，優先減少同時發生的動作，再逐步增加光線、質感或風格描述。需要參考時，可以看看Gemini 視頻提示詞中的現成範例。
哪些 AI 視頻模型支援圖片轉視頻
模型選單只顯示目前可用且支援圖片輸入的選項。不同模型的控制項可能不同，可依需要的動態效果與控制程度選擇。也可以在AI 視頻生成器中集中比較這些選項。
可以調整長寬比質素和聲音嗎
取決於所選視頻模型。頁面只展示目前模型支援的控制項，可能包括長寬比、質素、參考模式和聲音。如果某個選項沒有出現，或被鎖定為自動，表示目前模型或模式不支援手動覆蓋；切換模型後，可用設置也會隨之變化。
怎樣提高視頻中的主體一致性
使用主體清晰、細節完整的圖片，並讓提示詞圍繞一個主要動作展開。不要同時要求大幅改變姿勢、臉部、服裝和背景。如果主體開始失真，可以減少運動幅度、縮短提示詞，或在保留關鍵外觀描述的前提下更換模型進行比較。
Seedance 2.0 可以用於圖片轉視頻嗎
可以。Seedance 2.0 可在模型選單中用於圖片轉視頻，適合需要明確動作指令和電影感場景的創作。
Veo 3.1 可以用於圖片轉視頻嗎
可以。模型選單提供 Veo 3.1 時即可用於圖片轉視頻，適合追求精緻動態與場景連續性的創作。
