APIMODELSAPIMODELS

LTX 2.5 风格化与动画提示词

收录 80 多条 LTX-2.5 视频提示词,来自 Civitai、fal.ai 的 LTX-2.5 示例页和 X。每条都逐字全文收录、署名作者、链回原帖,原帖有视频的,卡片里直接播放。这些提示词都不是我们写的。Civitai 上的大多是作者在自己电脑上用 LTX-2.5 开源权重跑的(ComfyUI 里的 dev 或蒸馏版),模型是同一个家族,但没有经过我们的 API。

← 回到全部 89 条 LTX 2.5 提示词

最近更新: · LTX 2.5 Fast API 价格与在线体验 · API 文档

一句话摘要

LTX-2.5 是 Lightricks 在 2026 年 8 月 11 日发布的开源权重视频模型,每条视频都会连声音一起生成。apimodels.app 上有两档:LTX 2.5 Fast(ltx-2.5-fast)可以只给提示词,也可以再加一张首帧图或一段音频,输出 720p、1080p 或 1440p,24 到 50 fps,6 到 20 秒,$0.081/秒起;LTX 2.5 Pro(ltx-2.5-pro)是 Lightricks 定位给成片用的一档,只做文生和首帧图生,720p 或 1080p,6、8、10 秒三档,$0.108/秒起。提示词最长 5000 字符;没有 4K、尾帧、参考图和视频输入;失败不扣费。本页收录 80 多条别人公开发布的 LTX-2.5 提示词,来自 Civitai、fal.ai 的 LTX-2.5 页面和 X。Civitai 上很多是在本地用开源权重、配合作者自己的参数跑的,同一段文字放到 API 上,出来的效果可能不一样。

出处:Lightricks LTX(模型方官方站点)

悬停预览 · 点击有声

电梯卡通小品动画

A classic ink and grey watercolour newspaper gag cartoon comes to life with small, simple cartoon motion. The camera is fixed on a tripod and does not move, drift, zoom or push in at any point, so the framing at the end of the clip is exactly the framing at the start. The baffled office worker with the big nose and round glasses stays standing in the same spot holding his framed certificate against his chest the whole time: his eyes dart nervously back and forth between the up button and the down button on the wall panel, he leans a little toward the buttons and back again, his raised index finger hovers toward the panel and then pulls back without pressing anything, sweat drops fly off his forehead, and the question mark above his head bobs and wobbles. The elevator doors stay closed. The certificate text, the caption text at the bottom and every letter stay exactly as drawn and never change. Ink linework, crosshatching and flat grey wash are preserved exactly as drawn for the entire clip, with no photorealistic texture, no style drift and no added people, and he never walks or leaves the frame.

来自 Azimondious看原帖
悬停预览 · 点击有声

码头海军女兵

This video is shot in RSNature style animation with defined yet rough sketch lines, saturated painterly colors, and concept art draft quality. This style remains consistent across movement and scene changes. The video opens with a high-angle shot of some docks. A young woman in a Navy uniform stands leaning against a parapet. The camera moves closer, transitioning into a close-up just as the character turns toward the viewer. Her expression changes: surprise gives way to an ironic, knowing look. audio: Russian speech

来自 witalpro665看原帖
悬停预览 · 点击有声

平面动画情侣拥抱

hatching, Cute_niji_style, Flat colors, colored, 2d, thick outlines, 1man, 1woman, thick thighs,full body, blushing, smiling, standing, cuddling, hugging,man, man with beard and mustache, dark hair, hearts, arms around neck, arms around waist, kissing,

来自 Le_Grand_Duc看原帖
悬停预览 · 点击有声

暗黑动画霓虹小巷

A medium-wide shot in a 2D cel-shaded dark seinen anime style — fine inked outlines crisp and consistent throughout character and environmental edges, hard flat cel shadows with no gradient blending consuming the majority of the frame, dual neon colour bleed with vivid crimson-red pressing in from one side and cool electric blue-cyan from the opposing side, deep muted base skin and fabric tones overwhelmed by neon bleed, ink cross-hatch mark texture across clothing and secondary surfaces, distressed crumbling brick textures lining the alley walls, colour palette locked to near-black, vivid crimson, electric blue-cyan, and trace amber only. A narrow back alley stretches toward the camera, its far end open, crumbling brick walls running along both sides, deep shadow pools between them. A woman in her late twenties walks forward along the alley toward the camera in measured, deliberate steps — her sleek dark-brown bun tight against her head, grey-green eyes wide and scanning left and right beneath drawn brows, jaw clenched hard, her expression is scared and worried; she holds a houndstooth handbag pressed tight against her chest with both arms, a brown leather tote gripped at her side with white-knuckled fingers, shoulders pulled up high; she is dressed in a dark olive fitted blazer over a grey collared blouse, a knee-length muted teal-and-grey plaid tartan pencil skirt, and dark low-heeled pumps clicking against cracked stone pavement with each careful step. As she walks forward, the sharp protruding corner of a building gradually enters the frame ahead of her to one side — the edge of a building that marks a junction where a perpendicular side passage opens up beyond it. The camera holds a steady medium-wide framing and tracks gently backward in front of her, keeping her centred, a subtle persistent hand-camera tremor present throughout. She reaches the building corner, stops short of it, and stands upright on her own two feet; she extends her head and upper body slowly forward past the edge of the building corner, peering cautiously into the side passage beyond — scanning what lies around the turn; she holds there for a brief suspended beat, entirely still, breath held. Then she steps forward, turns the corner at the building edge, and walks away into the side passage, her figure receding and disappearing from view as she moves away. Audio: the measured click of low heels on cracked stone pavement echoing between brick walls, a distant drip of water, the faint ambient hum of an urban night beyond the alley, a shallow breath drawn and held as she peers around the corner, then the gradual fade of receding footsteps as she moves away into the passage — no music.

悬停预览 · 点击有声

全息霓虹音乐工作室

A pulsating, shimmering, shrinking and expanding, rotating, revolving, turning holographic ultraviolet music studio with a few neon lights in the corners and splotches of a diverse variety of neon paints on the walls, ceiling, and carpet with a variety of diverse electric guitars hanging on the right wall, a variety of diverse bass guitars hanging on the back wall, a variety of diverse keyboards hanging on the left wall, a variety of diverse woodwind musical instruments hanging from the ceiling, a variety of diverse stringed instruments on clear display stands on the floor. An old blind grizzled blues musician is playing a theremin on the left wall, an old blind grizzled jazz musician is playing drums on the right wall, an old blind grizzled rock and roll musician is playing a tuba on the ceiling, an old blind grizzled punk musician is playing the blues on a harmonican and an old blind grizzled rap musician is holding a skinny microphone and raps in a high nasally Jamaican accent: "thgplmnbghhhdfgh hhghhqqtrxxzmnwwslk jjnwwllktrh mmkkkkrrthppghbmnvcrtqwlkjggjfkk bmhlppqzzxmmqlpfgnnbvbc qqmwwrttmnzxdddfhrtpbbcn qlkcmnbzzwlk lwrthmbbkzqlp bmzzqmnc vvvvvmnnnq pprrtm pmnvvrt zlplwwrmngrnbvnwrv qlkmnnnvczzwpl?" as what he says appear as subtitles on the bottom of the screen in yellow letters with black outlines as the jazz musician plays jazz music, the blues musician plays blues music, the rock and roll musician plays rock and roll music, the punk musician plays punk music.

来自 msavitt623看原帖
悬停预览 · 点击有声

飞艇舱内的獾与鹤

Ten seconds, vertical 9:16, stylized 3D animation at feature quality, one continuous shot, ordinary speed. Setting: the rigging shaft running up the inside of a stricken airship at night. Riveted brass ribs, a ladder up one side, gas hoses crossing the space, a tear in the hull at the top of frame with cloud streaming through it. Emergency lamps throwing red. Cast: a stocky badger engineer in oiled canvas overalls, heavy fur, a wrench on a lanyard at his hip. A lean crane bird navigator in a cracked leather flight cap, long legs braced across two ribs. 0 to 3 seconds: the shaft lurches and drops. The badger comes off the ladder, catches a gas hose one-handed and swings out across the shaft. The camera tracks up with him from below. He shouts across: "The forward cell is gone, pull the ballast." 3 to 6 seconds: the bird throws his weight onto a brass release lever with both feet. The lever gives in two stages, the second further than the first. A column of ballast water dumps down past camera and out of frame. His beak stays shut through the whole effort. 6 to 8 seconds: the water clears. The badger lands back on the ladder, hooks an elbow through a rung, and the shaft steadies under them. The bird answers, flat: "That was the last of it." 8 to 10 seconds: both look up at the tear in the hull. Neither speaks. The camera keeps climbing past them toward the cloud coming through, and holds. Mouth rule: one mouth at a time. The badger's jaw moves only on his line and the bird's beak only on his, each holding still while the other talks. Through the final block neither mouth moves. Camera: one continuous climb up the shaft from low to high, accelerating slightly on the water dump and settling at the end. No cuts. Hold across the take: both characters, the overalls, the flight cap, the wrench on its lanyard, the rib spacing and the positions of the red lamps. Sound: two clearly separate voices, the badger low and strained under effort, the bird dry and thin. Structural groaning underneath the whole take, the hose creaking under load, the lever ratcheting in two stages, ballast water roaring past camera and falling away, wind through the tear. Nothing musical. Look: feature-quality stylized 3D, fur and feather detail holding through the motion, soft red emergency light against cold blue coming through the hull tear, real weight in the canvas overalls, water carrying volume. Do not: cut, ramp the speed, add dialogue beyond the two lines above, move both mouths together, or drop the wrench.

来自 fal.ai看原帖

怎么写好 LTX 2.5 提示词

  1. 1一个镜头写成一段话,按发生顺序写. 本页 fal.ai 的那些示例,每条都是从头到尾描述一个机位:景别和主体、发生了什么、镜头怎么动、光线,最后才是镜头焦段。大意是「缓慢推近擂台上的拳击手,顶光穿过烟雾打在肩上,50mm、f/2」。一串关键词堆在一起,模型拿不到先后顺序。
  2. 2把声音写进去,反正每条都会出声. LTX-2.5 的声音和画面一起生成,我们的 API 也没有静音选项,所以要不要声音不由你,怎么响可以由你。台词放进引号、写明是谁说的,再写环境音和最关键的那一下声音。本页有几条 Civitai 提示词把两个人的对话一句一句写好,每个人开口前先交代嗓音。
  3. 3动作长度要对得上 API 的时长档. Fast 的文生和图生是 6 到 20 秒、只能取偶数秒,1440p 或 48 / 50 fps 时只有 6、8、10 秒;Pro 只有 6、8、10 秒。本地工作流常用帧数定长度,除以帧率就知道原帖要的是几秒,再挑最接近的一档。API 不会看提示词自己定长度,要自己传 duration。
  4. 4传了首帧,就写动作,别再描述画面. 图生视频里,主体、场景和构图已经由那张图定死了,提示词就专心写什么在动:谁转身、镜头怎么走、听到什么。成片会保持图片的比例;带首帧又传 aspect_ratio,会直接返回 400。
  5. 5音频生视频:音轨说了算,提示词负责选角. LTX 2.5 Fast 收到 audio_url 后,会把这段音频原样当成片的声音,再按它生成画面。提示词写清楚画面里是谁、怎么动、怎么取景就够了,比如我们文档里的例子:明亮的演播室里,一位年轻主持人对着镜头讲话,手势自然,中景。长度默认跟着音频走,也可以自己设 duration。Pro 不收音频。
  6. 6API 收不了的东西,先删掉. 本地 LTX-2.5 工作流里可以挂负面提示词、LoRA 触发词、额外的引导图或尾帧。我们的请求只收提示词、一张首帧、一段音频和生成参数,别的都不收:尾帧、参考图、视频输入会返回 400。也没有负面提示词这个字段,原帖写在负面提示词里的东西,不挪进正文就等于没了。
  7. 7画面文字和很乱的物理,别指望一次成. Lightricks 自己的提示词指南也说了,画面里的文字和混乱的物理运动目前还不稳定。如果一个镜头靠一块看得清的招牌,或者一堆东西互相撞,做好多跑几次的准备,文字也可以剪辑时再加。

提示词公式

景别和主体 + 按先后写的一个清楚动作 + 镜头运动 + 光线和时间 + 焦段或胶片质感 + 引号里的台词和环境音 + 这一档允许的时长(用 duration 传,不写进提示词)

例(本库收录的原文,原样引用,作者 hichuicomfy,发在 Civitai):Style: high-resolution food photography - close-up - The chocolate is poured over the whipped cream and berries. The chocolate flows slowly, creating rich, glossy streaks across the white cream and the blue blueberry, the deep brown contrasting with the vibrant fruit. A soft, rich sound of chocolate dripping and settling accompanies the pour.

常见问题

LTX-2.5 是什么?

LTX-2.5 是 Lightricks 的视频模型,2026 年 8 月 11 日以开源权重发布,每条视频都会把声音和画面一起生成。Lightricks 的 API 分两档:Fast 主打速度和成本,Pro 按它的说法主打画面保真和运动稳定。Lightricks 还说 LTX-2.5 新增了原生多镜头、新的视频解码器和更好的提示词遵循,这些是厂商的说法,不是我们的测试结论。

这些提示词是谁写的?

都是别人写的:每一条都是创作者在 Civitai 或 X 上公开发的,或者是 fal.ai 放在它 LTX-2.5 页面上的,我们一个字都没写、也没改。每张卡片都逐字全文引用提示词、写明作者、链回原帖。提示词和视频归原作者所有;我们自己写的说明和整理的数据按 CC BY 4.0 授权。要更正署名或撤下某一条,发邮件给我们,或者在 GitHub 仓库提 issue。

Civitai 上的提示词,我用 API 跑出来不一样,为什么?

因为本页 Civitai 的提示词大多是作者在自己电脑上用 LTX-2.5 开源权重跑的,不是在我们的 API 上,同一段文字出来的效果自然会有差别。本地 ComfyUI 可能用的是 dev 或蒸馏版权重,还可能挂了 LoRA、调了采样参数、写了负面提示词或者加了引导图,这些都不在提示词文字里。文字能照搬,这些设置搬不过来。卡片上的视频是作者那套设置的结果,在 Fast 或 Pro 上要多试几次才能接近。

在哪能跑这些提示词?多少钱?

在 apimodels.app 上,模型名 ltx-2.5-fast 和 ltx-2.5-pro,接口 POST /v1/video/generations,按成片秒数计费。Fast:720p $0.081/秒、1080p $0.117/秒、1440p $0.171/秒,720p 6 秒 $0.486,1080p 10 秒 $1.17,1080p 20 秒 $2.34。Pro:720p $0.108/秒、1080p $0.153/秒,1080p 10 秒 $1.53。两档都比 Lightricks 官方 API 价低 10%;Fast 的音频生视频按成片长度计到 0.01 秒;失败不扣费。常见的用法是先在 Fast 上试提示词,选中的镜头再拿到 Pro 上出,两档字段一样,只改一个模型名。

本页的提示词放到 API 上被拒或者失败,为什么?

超出 API 限制的请求,会在建任务之前直接返回 400,不扣费。限制有这些:提示词最长 5000 字符;Fast 是 6 到 20 秒的偶数秒,Pro 只有 6、8、10 秒;没有 4K、尾帧、参考图和视频输入;Pro 不收音频、没有 1440p;传了首帧就不能再传 aspect_ratio。任务开始以后,也可能因为内容审核或上游繁忙而失败,同样不扣费。

LTX-2.5 会自己配声音吗?

会:每条视频的声音都和画面一起生成,我们的 API 也没有静音选项。在意声音的话,把台词放进引号,写明环境音。用 LTX 2.5 Fast 还可以直接传你自己的音轨(audio_url),画面会按这段音频来生成。

这些提示词和视频能商用吗?

提示词和视频归原作者所有,商用前先看原帖怎么说。你自己生成的视频,按你用来生成的那个服务的条款执行。如果你在本地跑开源权重,它用的是 Lightricks 的 LTX-2.x Community License,有自己的条件,商用前先读一遍。

在 apimodels.app 上跑这些提示词

LTX 2.5 Fast 已经上线,和另外 180+ 个图像、视频、语音、大模型共用一个 API、一个 key,按量计费、注册送 $0.10 试用额度。

这份提示词库在 GitHub 上是开源的

整份合集以 Markdown 和 JSON 放在 GitHub 上:每条提示词都带作者署名和原帖链接(Civitai、fal.ai 或 X)。提示词逐字全文引用,归原作者所有。视频也归作者所有,只链原帖、不转存。我们自己写的说明和整理的数据按 CC BY 4.0 授权。要更正署名或撤下某一条,提 issue 即可。

在 GitHub 上查看

其它提示词库