happyhorse-1.0-r2v

由 Alibaba 提供视频

通过 HappyHorse 1.0 参考图接口引导生成 720P 或 1080P 视频。

更新于:
输入提示词文本
模型7款可用
返回方式异步结果链接
价格起价 8.4quota / 秒
HappyHorse 1.0参考素材生成视频 结果
教室动作连锁反应

下方可对照 1.0 和 1.1 使用相同素材的结果。

参考素材生成视频 · 1.0
生成示例

相同输入,两版 HappyHorse 结果

01 · R2V

教室动作连锁反应

相同输入,两版结果
共用参考素材1 image
Female student action-scene reference
图像 1
HappyHorse 1.0当前模型
当前版本
HappyHorse 1.1对比版本
共用提示词两款模型使用相同的提示词

【Core Logic: Cause & Effect Chain】 1. Initiation: Protagonist (Female Student) is cornered. Opponent 1 lunges with a right punch. 2. Reaction & Momentum: Protagonist does NOT just dodge; she uses the opponent's momentum. She sidesteps left, grabbing Opponent 1's wrist to pull him off-balance. Her body rotates with the pull, using centrifugal force. 3. Environmental Interaction: As she spins, her foot kicks a loose chair leg. The chair slides across the floor (friction visible), tripping Opponent 2 who was rushing in. 4. Flow Transition: The spin continues. She releases Opponent 1, who crashes into a desk (wood splinters, papers fly). Her momentum carries her forward over the desk surface (parkour vault). 5. Landing & Reset: She lands on the other side of the desk in a crouch, absorbing impact. Opponent 3 hesitates. She rises smoothly, ready for the next engagement. Note: Every movement must have a clear physical cause. No floating, no sudden stops without deceleration. 【Visual Style & Camera Sync】Style: Gritty Realism, High-Budget Action Cinema. Natural light, dust motes, sweat, worn textures. Camera: "Physical Participant" perspective. ● When she pulls Opponent 1, camera whips LEFT to follow the rotation. ● When the chair slides, camera tilts DOWN to track the obstacle. ● When she vaults, camera BOOMS UP rapidly to match her vertical leap. ● Handheld shake matches the intensity of impacts. ● Foreground obstruction: Desks and arms frequently block the lens briefly to enhance depth and chaos. 【Character & Environment Details】Protagonist: Agile, wearing standard Japanese high school uniform (blazer/skirt or tracksuit), hair tied back but loosening. Movements are grounded, feet always interacting with floor friction. Opponents: Three aggressive males, reacting realistically to pain and balance loss (stumbling, flailing arms). Environment: Cluttered classroom. Desks are heavy (don't fly like cardboard). Papers swirl in turbulent air currents caused by movement. Sunlight creates dynamic shadows that move with the characters. 【Negative Prompts - Logic Specific】No teleporting, no instant direction changes without weight shift, no floating limbs, no enemies standing still after being hit, no magical effects, no slow-motion unless specified, no static background, no repetitive loops, no distorted anatomy during rapid motion, no clipping through objects.

02 · R2V

公主与龙人的对话

相同输入,两版结果
共用参考素材2 images
Dragon-man character reference
图像 1
Princess character reference
图像 2
HappyHorse 1.0当前模型
当前版本
HappyHorse 1.1对比版本
共用提示词两款模型使用相同的提示词

The princess [Image 2] was imprisoned in bedroom and be threatened by the dragon man [Image 1]. They had a long dialogue.

03 · R2V

直播服装展示

相同输入,两版结果
共用参考素材4 images
Livestream shopping scene reference
图像 1
Apparel outfit reference
图像 2
Pants product reference
图像 3
Livestream host reference
图像 4
HappyHorse 1.0当前模型
当前版本
HappyHorse 1.1对比版本
共用提示词两款模型使用相同的提示词

A female livestream host sells products from home [Image 4], using the livestream shopping scene as reference [Image 1]. The host speaks in an excited and energetic tone, holding up the pants from [Image 3] with her left hand. [0-4s] “Okay, be honest... how many sweatpants do you own that actually make you feel put-together?” After this line, the camera pushes in to a close-up of her fingers stretching the fabric, with a hook-like questioning tone. [5-9s] Cut to the host at home [Image 4], already changed into the outfit shown in [Image 2]. The camera pulls back to a half-body shot as she touches the fabric around her waist to demonstrate the material, speaking in an expressive tone. Line: “This isn't just loungewear. It's butter-soft ribbed knit, four-way stretch, high-waisted, stays put, and never rolls down.” [10-12s] Full-body shot showing the fit. Line: “Watch this. From a grocery run to a coffee date.” [12-15s] With a confident tone: “Link in bio. Grab yours before it sells out again.”

04 · R2V

霓虹夜景下的公寓对话

相同输入,两版结果
共用参考素材3 images
Woman character reference one
图像 1
Man character reference
图像 2
Woman character reference two
图像 3
HappyHorse 1.0当前模型
当前版本
HappyHorse 1.1对比版本
共用提示词两款模型使用相同的提示词

A modern minimalist living room at dusk, with a blurred city night view and neon bokeh outside the floor-to-ceiling window. In front of a dark gray fabric sofa is a low wooden coffee table with two glasses of water, a phone placed screen-down, and an open magazine. The woman from [Image 1], [Image 3], gently pushes a glass of water with her fingertips and says softly, “That day, when you told me to wait until the wind stopped, I believed you.” Cut to a close-up of the man from [Image 2]. His gaze darkens slightly: “I was the one who let go first.” The woman raises her eyes and looks straight ahead, her tone softening: “So you even skipped the apology?” The two never make eye contact. The camera pushes in to an over-the-shoulder side profile of Lin Wan. The corner of her mouth lifts slightly, but the smile never reaches her eyes: “Leave? You taught me how to leave, but never gave me the address of the door.” The woman walks to the floor-to-ceiling window, while the man remains seated on the sofa, head lowered in silence. The cool blue neon outside blends naturally with the warm interior light. The scene maintains cinematic shallow depth of field and subtle film grain in a 16:9 frame, with lip sync aligned to dialogue, restrained pacing, and a realistic short-drama look.

05 · R2V

酒吧里的牛仔对话

相同输入,两版结果
共用参考素材2 images
First cowboy character reference
图像 1
Second cowboy character reference
图像 2
HappyHorse 1.0当前模型
当前版本
HappyHorse 1.1对比版本
共用提示词两款模型使用相同的提示词

Two cowboys, [Image 1] and [Image 2], are drinking in a bar and talking about their future. The two have an extended dialogue.

模型能力

happyhorse-1.0-r2v 能做什么

根据提示词生成

提交简洁提示词,通过一个 TTAPI 任务获取视频结果。

常用参数

设置提示词、模型、输出参数和回调即可接入。

异步查询

通过轮询或 Webhook 接收结果,避免耗时生成阻塞界面。

接入产品

HappyHorse 的输出可以展示、保存或交给后续流程处理。

计费

按操作计费,单位为 quota

按模型、操作和输出设置扣除 quota。

视频生成官方定价
分辨率价格计费单位
720P8.4 quota / 秒
1080P14.4 quota / 秒
API 请求

请求头与提交参数

使用 TTAPI 密钥,按下方字段向接口发送请求。

提交任务POST
/happyhorse/api/v1/services/aigc/video-generation/reference-to-video

提交 HappyHorse 参考素材生成视频任务。

查看接口文档
查询结果GET
/happyhorse/api/v1/tasks/{task_id}

取回 HappyHorse 任务结果。

查看接口文档
POSThttps://api.ttapi.io/happyhorse/api/v1/services/aigc/video-generation/reference-to-video

请求头

TT-API-KEYstring必填

填写 TTAPI API 密钥。

Content-Typestring必填

使用 application/json。

请求体

modelstring必填

HappyHorse 任务的模型 ID,例如 happyhorse-1.1-t2v。

input.promptstring按条件填写

描述生成或编辑内容的提示词。

input.mediaobject[]按条件填写

按所选操作传入首帧图、参考图或源视频。

parameters.resolutionenum选填

输出分辨率为 720P 或 1080P。

parameters.durationinteger选填

视频时长,单位为秒。

parameters.ratioenum选填

所选工作流支持的输出宽高比。

parameters.watermarkboolean选填

是否添加 HappyHorse 水印。

parameters.seedinteger选填

随机种子,范围 0 至 2147483647。

查看 TTAPI 接入文档
接入示例

发送请求并获取结果

复制当前模型的请求地址、鉴权请求头和请求体。

curl --request POST \
  --url https://api.ttapi.io/happyhorse/api/v1/services/aigc/video-generation/reference-to-video \
  --header "TT-API-KEY: $TTAPI_KEY" \
  --header 'Content-Type: application/json' \
  --data '{"model":"happyhorse-1.0-r2v","input":{"prompt":"slow dolly shot through a lantern-lit alley","media":[{"type":"reference_image","url":"https://example.com/reference.jpg"}]},"parameters":{"resolution":"720P","duration":5,"watermark":false}}'