使用 RunComfy 的 API 运行 minimax/minimax-h3-max/reference-to-video。 可接受的输入与输出请参阅模型的 数据结构说明。
curl --request POST \
--url https://model-api.runcomfy.net/v1/models/minimax/minimax-h3-max/reference-to-video \
--header "Content-Type: application/json" \
--header "Authorization: Bearer <token>" \
--data '{
"prompt": "Image 1 is the near man; Image 2 is the far man. Keep both faces, hair, and jackets identical to their reference images. Dusk in an empty asphalt parking lot under soft sodium lights. They stand a few meters apart. The near man says flatly: \"So… you actually tried it?\" A beat of wind. The far man answers with a slow half-smile, warm and certain: \"Yeah. I love it.\" The near man exhales and shakes his head once, almost smiling. Static camera, medium wide. Native audio, lip-synced English dialogue, distant traffic hum, wind. No music, no subtitles, no camera movement, no text, no watermark.",
"prompt_expansion_mode": "balanced",
"reference_images": [
"https://playgrounds-storage-public.runcomfy.net/tools/7453/media-files/ref-1.webp",
"https://playgrounds-storage-public.runcomfy.net/tools/7453/media-files/ref-2.webp"
]
}'将 YOUR_API_TOKEN 环境变量设置为您的 API 密钥(在 个人资料中管理密钥),并在每个请求的 Authorization 标头中以 Bearer 令牌形式携带: Authorization: Bearer $YOUR_API_TOKEN。
提交异步生成任务,将立即获得 request_id 以及用于查询状态、获取结果和取消的 URL。
curl --request POST \
--url https://model-api.runcomfy.net/v1/models/minimax/minimax-h3-max/reference-to-video \
--header "Content-Type: application/json" \
--header "Authorization: Bearer <token>" \
--data '{
"prompt": "Image 1 is the near man; Image 2 is the far man. Keep both faces, hair, and jackets identical to their reference images. Dusk in an empty asphalt parking lot under soft sodium lights. They stand a few meters apart. The near man says flatly: \"So… you actually tried it?\" A beat of wind. The far man answers with a slow half-smile, warm and certain: \"Yeah. I love it.\" The near man exhales and shakes his head once, almost smiling. Static camera, medium wide. Native audio, lip-synced English dialogue, distant traffic hum, wind. No music, no subtitles, no camera movement, no text, no watermark.",
"prompt_expansion_mode": "balanced",
"reference_images": [
"https://playgrounds-storage-public.runcomfy.net/tools/7453/media-files/ref-1.webp",
"https://playgrounds-storage-public.runcomfy.net/tools/7453/media-files/ref-2.webp"
]
}'根据 request_id 获取当前状态("in_queue"、"in_progress"、"completed" 或 "cancelled")。
curl --request GET \
--url https://model-api.runcomfy.net/v1/requests/{request_id}/status \
--header "Authorization: Bearer <token>"获取指定 request_id 的最终输出与元数据;若任务未完成,响应会返回当前状态,便于继续轮询。
curl --request GET \
--url https://model-api.runcomfy.net/v1/requests/{request_id}/result \
--header "Authorization: Bearer <token>"通过 request_id 取消排队中的任务;进行中的任务无法取消。
curl --request POST \
--url https://model-api.runcomfy.net/v1/requests/{request_id}/cancel \
--header "Authorization: Bearer <token>"请提供可公网访问的 HTTPS 地址。确保目标主机允许服务端拉取(无需登录或 Cookie)、未被限流或拦截爬虫。建议:图片 ≤ 50 MB(约 4K),视频 ≤ 100 MB(约 720p 下 2–5 分钟)。私有资源请使用稳定或预签名 URL。
{
"type": "object",
"title": "输入结构",
"required": [
"prompt",
"prompt_expansion_mode",
"reference_images"
],
"properties": {
"prompt": {
"title": "提示词",
"description": "描述场景、动作、摄像机和音频。将资源称为图像 1、图像 2、视频 1、音频 1,并说明每个参考提供的内容。",
"type": "string",
"default": "Image 1 is the near man; Image 2 is the far man. Keep both faces, hair, and jackets identical to their reference images. Dusk in an empty asphalt parking lot under soft sodium lights. They stand a few meters apart. The near man says flatly: \"So… you actually tried it?\" A beat of wind. The far man answers with a slow half-smile, warm and certain: \"Yeah. I love it.\" The near man exhales and shakes his head once, almost smiling. Static camera, medium wide. Native audio, lip-synced English dialogue, distant traffic hum, wind. No music, no subtitles, no camera movement, no text, no watermark."
},
"reference_images": {
"title": "参考图片",
"description": "主题或样式参考图像 URL,在提示中引用为图像 1、图像 2 等。结合视频和音频,最多 12 个参考文件。",
"type": "array",
"default": [
"https://playgrounds-storage-public.runcomfy.net/tools/7453/media-files/ref-1.webp",
"https://playgrounds-storage-public.runcomfy.net/tools/7453/media-files/ref-2.webp"
],
"items": {
"type": "string",
"format": "image_uri"
},
"maxItems": 12
},
"reference_videos": {
"title": "参考视频",
"description": "运动参考片段(每个约2-15秒,组合时长最多15秒),引用为视频1、视频2。结合图像和音频,最多12个参考文件。",
"type": "array",
"items": {
"type": "string",
"format": "video_uri"
},
"maxItems": 12
},
"reference_audios": {
"title": "参考音频",
"description": "可选的语音或氛围片段(每个约 2-15 秒;组合持续时间最多 15 秒)。不能作为唯一参考;至少提供一张图片或视频。",
"type": "array",
"items": {
"type": "string",
"format": "audio_uri"
},
"maxItems": 12
},
"aspect_ratio": {
"title": "宽高比",
"description": "输出纵横比。 adaptive 尽可能遵循参考框架。",
"type": "string",
"enum": [
"adaptive",
"21:9",
"16:9",
"4:3",
"1:1",
"3:4",
"9:16"
],
"default": "adaptive"
},
"resolution": {
"title": "分辨率",
"description": "本机生成分辨率。 480p 更快/更便宜; 768p 是更清晰的默认值。",
"type": "string",
"enum": [
"480p",
"768p"
],
"default": "768p"
},
"duration": {
"title": "时长",
"description": "Length of the generated video in seconds (5-15).",
"type": "integer",
"minimum": 5,
"maximum": 15,
"default": 5
},
"prompt_expansion_mode": {
"title": "提示词扩展模式",
"description": "生成前要花多少精力重写提示。 balanced 返回很快; quality 在更丰富的提示上花费更长的时间。",
"type": "string",
"enum": [
"balanced",
"quality"
],
"default": "balanced"
},
"seed": {
"title": "种子",
"description": "固定种子以获得可重复的结果。使用 -1 作为随机种子。",
"type": "integer",
"default": -1
},
"enable_safety_checker": {
"title": "启用安全检查",
"description": "如果设置为 true,将启用安全检查器。",
"type": "boolean",
"default": true
}
}
}{
"output": {
"type": "object",
"properties": {
"image": {
"type": "string",
"format": "uri",
"description": "单张图片 URL"
},
"video": {
"type": "string",
"format": "uri",
"description": "单个视频 URL"
},
"images": {
"type": "array",
"description": "多张图片 URL",
"items": {
"type": "string",
"format": "uri"
}
},
"videos": {
"type": "array",
"description": "多个视频 URL",
"items": {
"type": "string",
"format": "uri"
}
}
}
}
}RunComfy 是首选的 ComfyUI 平台,提供 ComfyUI 在线 环境和服务,以及 ComfyUI 工作流 具有惊艳的视觉效果。 RunComfy还提供 AI Models, 帮助艺术家利用最新的AI工具创作出令人惊叹的艺术作品。