Cinel AI
靈感庫
全部靈感瀏覽全部作品人像寫真產品與電商海報與平面設計資訊圖與 UI3D 角色與公仔美食飲品插畫與動漫創意場景
AI 圖片
免費 AI 圖片生成器從文字提示詞創作圖片文生圖生成器從文字提示詞創作圖片AI 圖生圖生成器根據參考圖生成圖片AI 圖片編輯器用文字編輯已有圖片背景移除即時移除圖片背景圖片放大圖片放大 2 倍或 4 倍AI 擴圖擴充套件圖片原有邊界
AI 影片
AI 影片生成器開啟影片生成工作臺文生影片生成器從文字提示詞創作影片圖生影片讓參考圖片動起來
AI 模型

圖片模型

FLUX.2 Klein 4BFLUX.2 Klein 極速生圖,支援提示詞與參考圖。KolorsKolors 免費生圖,支援提示詞與參考圖。Seedream 4.0低成本 1K、2K、4K 生成,支援多參考圖編輯。Seedream 4.54K 生成、文字渲染與參考圖編輯。Seedream 5.0 Lite電影感 3K 寫實,支援多參考圖創意控制。Seedream 5 Pro提供 1K 快速、1.5K 輕量和 2K 專業視覺,支援文字與多參考圖控制。GPT Image 2精準理解提示詞,文字清晰,適合產品視覺。GPT Image 2.5更快的 GPT 影象生成,編輯更準,主體更穩定。Nano Banana 2 Lite快速 1K 生成與編輯,適合高頻迭代。Nano Banana 2多參考圖編輯與快速風格控制。Nano Banana Pro4K 渲染與高階影象編輯。Nano Banana輕量 Gemini 編輯,適合快速修改。Nano Banana 2.1最新一代 Nano Banana 生成與多參考圖編輯。

影片模型

Seedance 1.0 Turbo快速影片生成,動作流暢,最高 1080p。Seedance 2.54–30 秒影片,支援同步音訊、480p 或 720p。Seedance 2.0 Mini低成本 4–15 秒影片,支援同步音訊、480p 或 720p。Seedance 2.04–15 秒影片,支援 Fast/Standard,最高 1080p。MiniMax H3高質量 4–15 秒影片,支援首幀和多圖參考。Kling 3.0強動態、運鏡與可選聲音,最高 4K。Kling 3.0 Turbo更快的 3–15 秒影片,支援 720p、1080p 和首幀輸入。Kling O1Kling O1 圖片轉影片,支援首幀和可選尾幀。Kling O3Kling O3 文生影片與首尾幀生成,最高支援 4K。Grok Imagine Video 1.06–30 秒測試版影片,支援 Fun、Normal 和 Spicy 風格模式。Grok Imagine Video 1.51–15 秒影片,支援原生音訊和最多 7 張參考圖。Wan 2.55–10 秒文生或單圖生影片,支援原生音訊。Wan 2.62–15 秒多鏡頭影片,支援原生音訊和 720p、1080p。Wan 2.72–15 秒影片,支援文字、首尾幀和最多三張參考圖。Veo 3.1 Fast穩定版 Veo 3.1 Fast,支援 4–8 秒影片、音訊和首尾幀輸入。Veo 3.1 Pro穩定版 Veo 3.1 Pro,支援 4–8 秒影片、音訊和首尾幀輸入。Veo 3.1 Fast BetaFast Beta 固定 8 秒影片,支援首尾幀或三張參考圖。Veo 3.1 Pro BetaPro Beta 固定 8 秒影片,支援首尾幀輸入。
價格
  1. 靈感庫
  2. 創意場景
  3. 萬聖節攝像機骷髏鬥舞關鍵幀

萬聖節攝像機骷髏鬥舞關鍵幀

STYLE: Photoreal raw flash footage, like a found Halloween night tape. Shot on a cheap early-2000s consumer camcorder with a fisheye lens: strong barrel distortion, harsh on-camera flash/light blasting whatever is close to the lens with blown-out highlights and hard shadows, everything past a few meters falls off into an underexposed, near-black background. 

Heavy camcorder noise, video grain, slight interlacing, smeared motion blur, auto-exposure pumping, occasional focus hunting, slightly crushed and green-tinted digital color. Chaotic documentary realism: the operator is a person running and reacting in real time, frantic handheld whip-pans, crash zooms, the camera bumping into things, subjects shoved right up into the fisheye lens. Meme energy: absurd moments played totally real. 

FAST pace: hard jump cuts every 0.5-1 second, no transitions. No on-screen text, no timestamp overlay, no titles, no logos, no watermark. CAST: @image (15) the devil woman. 

Dale the werewolf: a huge, hulking, shaggy grey-brown werewolf with glowing yellow eyes, a long snarling snout full of teeth, and matted fur. He wears a torn red varsity-style bowling jacket hanging off his massive frame, shredded at the sleeves. The skeletons: a trio of full-body human skeletons with bleached off-white bones, wearing mirrored aviator sunglasses and black tactical vests. One of them wears a bright orange baseball cap and white sneakers. The elderly witch: a frail, wrinkled old woman with wild grey hair, wearing a black feathered cape and sharp cat-eye glasses, with a pointed witch's hat and a crooked grin. The Headless Horseman: a tall rider in a burgundy velvet coat with no head, seated on a powerful black horse with a flowing mane. Shot 1 (0-2.2s): Flash-lit fisheye shot at a gas station at night, the devil woman bored against an old pickup, camera close on her face; out of the black behind her, the werewolf sprints into the flash and leaps straight at the lens; she swings up a shotgun and fires, muzzle flash whiting out the frame, camera knocked sideways. Shot 2 (2.2-2.8s): Whip-pan into a dark alley: three skeletons on a low wall lit only by the flash, one raises his rifle and they jump down at the camera. Shot 3 (2.8-3.6s): The devil woman leaps out of a black archway right past the lens, her red tulle skirt and sheer sleeves flaring and filling the frame, firing a rifle. Shot 4 (3.6-4s): Flash-blown close-up of muzzle flash and sparks, camera shaking. Shot 5 (4-5s): Running handheld through a dark farmyard toward a shed with propane tanks; the elderly witch zooms past the lens on a ride-on lawnmower, cape flapping, straight into the tanks. 

Shot 6 (5-6s): Massive fireball overexposes the camera; the witch is blasted up into the black sky; operator falls back. Shot 7 (6-6.6s): Pitch black cornfield, flash pops on: her broom stabs into the dirt right in front of the lens. Shot 8 (6.6-7.8s): 

City street at night, camera on the sidewalk; the Headless Horseman gallops inches past the fisheye lens holding a flaming jack-o'-lantern, fire flaring the image. Shot 9 (7.8-8.8s): 

A parked old sedan explodes, camera recoils, image clips to white then auto-exposes back. Shot 10 (8.8-12.2s): Packed barn Halloween party, flash on, crowd in witch hats and bedsheet ghosts screaming and holding up phones. 

Dance battle filmed from the front row: the devil woman, in her full devil costume with horns and heels, hits a sharp freeze right in the fisheye lens; the skeleton in the orange cap and white sneakers does a headspin, then collapses into a pile of loose bones. 

Crowd goes insane, camera shaking. Shot 11 (12.2-14s): Camera drops to the floor: harsh flash on scattered bones, the skull rolls straight into the lens and stops, distorted by the fisheye, background black. Hold. AUDIO: Raw camcorder mic audio, clipping and distorted: wind, werewolf roar, shotgun blast, people yelling 'YO!', lawnmower engine, explosions blowing out the mic, hooves, party crowd screaming, bones clattering, skull rolling. No music, no score.
Photoreal raw flash camcorder video keyframe, vertical 9:16, like a found Halloween night tape: front-row fisheye view of a packed barn Halloween party, harsh on-camera flash blasting what is close to the lens, blown highlights and hard shadows, everything far away falls into near-black. A full-body skeleton character with bleached off-white bones, mirrored aviator sunglasses, a black vest, a bright orange baseball cap and plain white sneakers is frozen mid-headspin on the straw-strewn dance floor; behind him a crowd in witch hats and bedsheet ghost costumes screams and cheers. Strong barrel distortion, heavy camcorder noise and grain, slight interlacing, smeared motion blur, slightly crushed green-tinted digital colour, absurd moment played totally real. No on-screen text, no timestamp, no brand, no logo. Vertical 9:16 composition.
  • Rewrote the Seedance 2.5 video prompt as one vertical 9:16 key-frame still (kept the video's aspect); removed reference-image dependencies, brands, logos and celebrity/IP elements

cinel 為生成示例圖做了以上替換。

會員 1K 2 積分,每日積分 4。 比例 9:16。

ratio 9x16text only電影感寫實復古video sourced

引數

分類創意場景
畫面比例9:16
輸入純文字
示例圖模型GPT Image 2.5
推薦模型GPT Image 2, Nano Banana 2, Seedream 5 Pro
原作者@Just_sharon7 · X 原帖
原帖模型seedance-2.5

這條提示詞為什麼有效

根據 @Just_sharon7 於 2026-10-19 的影片帖改寫(記錄時 448 贊)。我們改成一張乾淨的9:16豎版圖,去掉品牌、Logo、名人/IP元素,也不需要你的照片。一次只改一個變數。

如何改寫

變數作用示例
frozen mid-headspin on the straw-strewn dance floor派對瞬間同一場景換個主體, 同一主體換個場景

使用提示

  • 從影片提示詞中挑一個高光瞬間,定格為單幀。
  • 一次只改一個變數,畫面才易讀。
  • 保留原影片的9:16豎版畫幅,並要求畫面無文字。

相關推薦

復古相機變身可居住建築1:1復古相機變身可居住建築@Gdgtify
極光冰梯浪漫主義油畫1:1極光冰梯浪漫主義油畫@NayaVerseee
廢棄教室對決餘波關鍵幀16:9廢棄教室對決餘波關鍵幀@Alina_with_Ai
發光蝸牛日落水面桌布1:1發光蝸牛日落水面桌布@churvikv
2003年DV攝像機商場冰淇淋約會關鍵幀16:92003年DV攝像機商場冰淇淋約會關鍵幀@AIwithkhan
不等電梯的不可能樓梯關鍵幀16:9不等電梯的不可能樓梯關鍵幀@umesh_ai
兒童友好物理漫畫資訊圖1:1兒童友好物理漫畫資訊圖@Gdgtify
小烏龜樹葉滑翔山谷關鍵幀16:9小烏龜樹葉滑翔山谷關鍵幀@AIwithWania
水粉泡泡字城市插畫海報1:1水粉泡泡字城市插畫海報@Goodmanprotocol
雨夜咖啡館熔岩巧克力蛋糕1:1雨夜咖啡館熔岩巧克力蛋糕@Nokiardnv
紙雕鯨魚天空立體場景1:1紙雕鯨魚天空立體場景@AllaAisling
行李箱變形飛行摩托關鍵幀16:9行李箱變形飛行摩托關鍵幀@Aiminithoughts
← 上一條復古相機變身可居住建築下一條 →廢棄教室對決餘波關鍵幀
1× · 1K
Cinel AI

Cinel 是免費的無限 AI 文生圖生成器,提供提示詞圖片編輯和 AI 圖生影片工具。無需註冊即可創作圖片,使用 GPT Image 2、Seedream 5 Pro 和 Nano Banana 2 編輯,再使用 Seedance 2.5、Veo 3.1 和 Kling 3.0 將靜態畫面變成影片。

產品免費 AI 圖片生成器AI 影片生成器圖片模板工作臺文生圖生成器圖生圖文生影片生成器圖生影片AI 圖片編輯背景移除
靈感庫全部靈感人像寫真產品與電商海報與平面設計資訊圖與 UI3D 角色與公仔美食飲品插畫與動漫創意場景GPT Image 2.5 提示詞Nano Banana 2.1 提示詞來源與下架

AI 圖片模型

FLUX.2 Klein 4BKolorsSeedream 4.0Seedream 4.5Seedream 5.0 LiteSeedream 5 ProGPT Image 2GPT Image 2.5Nano Banana 2 LiteNano Banana 2Nano Banana ProNano BananaNano Banana 2.1

AI 影片模型

Seedance 1.0 TurboSeedance 2.5Seedance 2.0 MiniSeedance 2.0MiniMax H3Kling 3.0Kling 3.0 TurboKling O1Kling O3Grok Imagine Video 1.0Grok Imagine Video 1.5Wan 2.5Wan 2.6Wan 2.7Veo 3.1 FastVeo 3.1 ProVeo 3.1 Fast BetaVeo 3.1 Pro BetaSora 2
公司價格關於我們聯絡我們support@cinel.ai隱私條款可接受使用政策
© 2026 Cinel AI。保留所有權利。