Froging AI
首頁

文字轉影片:用文字描述你的鏡頭

以文字描述人物、場景與動作,即可建立 AI 影片。選擇模型、片長及比例,完成後預覽並下載。

所需點數會依模型與設定而改變。

作品範例

Seedance 2.0

雨夜巷弄中的偵探

偵探走過濕漉漉的巷弄,在一盞孤燈下收起雨傘。

閱讀原始提示詞

Classic black-and-white noir cinema, high contrast silver gelatin texture. One adult female detective in a trench coat carries an umbrella through a narrow wet brick alley at night. She walks through one overhead pool of light, casting a long moving shadow on the brick wall as raindrops ripple in puddles. Sideways tracking camera keeps her silhouette clear. No weapons, no other people. Single continuous shot, exactly five seconds, no subtitles, no watermark.

Seedance 2.0

追逐風吹帽子的企鵝

圓滾滾的卡通企鵝在冰面追著黃色帽子,滑行後緩緩停下。

閱讀原始提示詞

Playful polished 3D cartoon animation, bright clean pastel arctic world. One round baby penguin waddles quickly after a small yellow knitted hat rolling across smooth pale ice in the wind. It reaches for the hat, misses by a little and slides to a gentle stop. Side tracking camera, expressive squash-and-stretch body but consistent two flippers and two feet. Clear comedic action. Single continuous shot, exactly five seconds, no subtitles, no watermark.

Seedance 2.0

穿越陽光河面的三匹馬

三匹馬涉過淺河,側向跟拍的鏡頭捕捉逆光下閃亮的水花。

閱讀原始提示詞

Photoreal wildlife documentary, wide side view at sunrise. Three chestnut horses trot from left to right through a shallow river, water reaching only their lower legs. Backlit water sprays sparkle against dark distant mountains, natural anatomy and coordinated strides. Camera smoothly tracks parallel, keep all three separate and fully visible, no riders, no cuts. Single continuous shot, exactly five seconds, no subtitles, no watermark.

Seedance 2.0

剝開溫熱可頌的瞬間

雙手剝開溫熱可頌,金黃色的酥皮碎片落在木板上。

閱讀原始提示詞

Warm authentic bakery food film, close-up of just two adult hands gently pulling apart one freshly baked croissant above a wooden board. The crisp golden shell breaks into flaky crumbs that fall onto the board, revealing soft airy layers and a small wisp of steam. Warm window side light, cream background, fixed macro camera. Anatomically correct hands, no extra fingers, no text. Single continuous shot, exactly five seconds, no subtitles, no watermark.

一個鏡頭,一個重點

先寫誰在做什麼,再補上地點、景別與光線。與其只說「電影感」,不如明確指定低角度跟拍或固定特寫。

同時安排畫面與聲音

支援音訊的模型可分別描述說話者、簡短台詞與環境音。不支援音訊時,先專注畫面,後續再加上音樂或旁白。

以單一鏡頭驗收

檢查文字中的主角、動作與視角是否出現在影片中。長故事應分鏡生成後再剪接;即使模型支援音訊,也要播放檢查台詞與聲畫同步。

常見問題

以文字描述人物、場景與動作,即可建立 AI 影片。選擇模型、片長及比例,完成後預覽並下載。

需要上傳圖片嗎?

不需要,只要用文字描述主體、動作及場景。若想讓現有照片動起來,請改用圖片轉影片。

短片提示詞應包含什麼?

先寫主體、動作、場景及鏡頭,再補充光線與風格。一個短片聚焦一件事,避免要求多次互不相關的轉場。

更換模型會改變設定嗎?

會。片長、解析度、比例、音訊選項與費用都依模型更新。即使用同一句提示詞比較,也要確認各模型輸出條件。

能一次生成完整長片嗎?

此工具製作的是模型片長限制內的獨立片段。較長作品請分鏡生成,下載後再用剪輯軟體串接。

相關內容