关于 video-editing
video-editing 是通过 each::sense AI Agent 和 each::labs API 编辑现有视频的技能。它将自然语言请求用于色彩分级、速度调整、画面增强、剪辑、特效和画幅重排,并可使用图像作为参考。
主要功能
- 01自然语言视频编辑请求
- 02色彩分级与视觉风格迁移
- 03速度调整与时间段指令
- 04画面增强、剪辑与特效
- 05宽高比调整及图像参考
适用场景
- 01为原始素材应用色彩分级
- 02放慢片段中的指定时刻
- 03根据参考图匹配视频风格
- 04将横屏素材改为竖屏社交媒体格式
安装
复制下面的命令,在终端中运行。
$
npx skills add https:/ / github.com/ awesome-genmedia/ skills --skill video-editing安装后在 Claude Code 中输入/video-editing确认已加载。
详细介绍
视频编辑
使用 each::sense(智能 AI Agent,可自动为你的请求选择最佳模型)通过 AI 驱动的工具编辑和转换现有视频。
快速开始
需要 each::labs API 密钥。请前往 eachlabs.ai 获取。
使用 curl
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"messages": [{"role": "user", "content": [
{"type": "text", "text": "Apply a warm cinematic color grade to this video. Boost the orange tones in highlights and teal in the shadows, add slight film grain."},
{"type": "image_url", "image_url": {"url": "https://example.com/raw-footage.mp4"}}
]
}
],
"stream": false
}'
使用 Python(OpenAI SDK)
from openai import OpenAI
client = OpenAI(
api_key="YOUR_EACHLABS_API_KEY",
base_url="https://eachsense-agent.core.eachlabs.run/v1"
)
response = client.chat.completions.create(
model="eachsense/beta",
messages=[{"role": "user", "content": "Apply a warm cinematic color grade to this video. Boost the orange tones in highlights and teal in the shadows, add slight film grain."}],
# Images are included in the message content array above
)
print(response.choices[0].message.content)
使用参考图像
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"messages": [{"role": "user", "content": [
{"type": "text", "text": "Apply the color grading and visual style from this reference image to my video clip"},
{"type": "image_url", "image_url": {"url": "https://example.com/my-video.mp4"}},
{"type": "image_url", "image_url": {"url": "https://example.com/style-reference.jpg"}}
]
}
],
"stream": false
}'
图像通过 OpenAI 多模态内容格式嵌入消息中发送。每个请求最多可包含 4 张图像。
流式传输
设置 "stream": true 以获取实时 SSE 响应,或设置 "stream": false" 以在单个响应中获取完整结果。流式传输适合在 UI 中显示进度;非流式传输则更适合脚本和自动化。
编辑能力
| 类别 | 可执行的操作 | 指令示例 |
|---|---|---|
| 调色 | 调整色彩、对比度和氛围 | "应用氛围感的青色与橙色调色" |
| 速度控制 | 慢动作、延时摄影、变速 | "转换为 0.5x 倍速的流畅慢动作" |
| 风格迁移 | 应用艺术风格 | "让画面看起来像一部 Wes Anderson 电影" |
| 画质增强 | 锐化、降噪、稳定画面 | "稳定这段抖动的手持拍摄素材" |
| 剪辑 | 剪切和选择片段 | "只保留此片段的前 5 秒" |
| 特效 | 添加视觉特效 | "在太阳出现时添加细微的镜头光晕" |
| 格式 | 更改宽高比、裁剪 | "裁剪为适用于 TikTok 的 9:16 竖屏,聚焦中央主体" |
| 转场 | 添加转场效果 | "在开头添加平滑淡入,在结尾添加淡出" |
提示词技巧
描述目标效果
具体描述你想要的视觉效果:
"Make this look like it was shot on 35mm Kodak Portra 400 film.
Desaturate slightly, add warm highlights, lift the blacks,
and include subtle film grain."
指定时间变化
当编辑需要应用于特定时刻时:
"Slow motion from 0:02 to 0:04 when the ball is in the air,
then return to normal speed."
基于参考的编辑
使用第二张图像或帧作为风格参考:
"Match the color grading of the reference image.
Apply the same contrast, saturation, and color temperature."
示例
电影感调色
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"messages": [{"role": "user", "content": [
{"type": "text", "text": "Apply a cinematic color grade inspired by Blade Runner 2049. Push shadows toward deep blue, highlights toward warm amber. High contrast, slightly desaturated midtones, add atmospheric haze effect."},
{"type": "image_url", "image_url": {"url": "https://example.com/city-footage.mp4"}}
]
}
],
"stream": false
}'
慢动作效果
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"messages": [{"role": "user", "content": [
{"type": "text", "text": "Convert this video to smooth slow motion at 0.25x speed. Use frame interpolation to keep the motion fluid and artifact-free. Maintain the original resolution."},
{"type": "image_url", "image_url": {"url": "https://example.com/action-clip.mp4"}}
]
}
],
"stream": false
}'
转换为动画风格
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"messages": [{"role": "user", "content": [
{"type": "text", "text": "Transform this real-world video into a Studio Ghibli anime style. Keep the motion and composition, but render everything as hand-painted animation with soft colors and visible brushstrokes."},
{"type": "image_url", "image_url": {"url": "https://example.com/nature-walk.mp4"}}
]
}
],
"stream": false
}'
社交媒体格式调整
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"messages": [{"role": "user", "content": [
{"type": "text", "text": "Reformat this 16:9 landscape video to 9:16 vertical for TikTok. Intelligently crop to keep the main subject centered. Add a quick zoom-in effect at the beginning for engagement."},
{"type": "image_url", "image_url": {"url": "https://example.com/horizontal-clip.mp4"}}
]
}
],
"stream": false
}'
复古胶片效果
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"messages": [{"role": "user", "content": [
{"type": "text", "text": "Make this video look like it was recorded on a VHS camcorder from the 1990s. Add scan lines, tracking artifacts, slight color bleeding, a date stamp in the corner reading 08/15/1994, and lower the resolution slightly."},
{"type": "image_url", "image_url": {"url": "https://example.com/modern-video.mp4"}}
]
}
],
"stream": false
}'
调色预设
可在提示词中使用的常见调色描述:
| 风格 | 描述 |
|---|---|
| 青色与橙色 | 冷色阴影、暖色高光,广受欢迎的大片风格 |
| 黑色电影 | 高对比度黑白效果、深邃阴影、明亮高光 |
| 复古胶片 | 提亮黑色、褪色高光、温暖柔和的色调 |
| 霓虹赛博朋克 | 高饱和度的洋红色与青色、高对比度、深黑色 |
| 粉彩梦境 | 降低饱和度、提亮阴影、柔和的粉色与蓝色色调 |
| 纪录片 | 自然色彩、轻微降低饱和度、中性对比度 |
| 黄金时刻 | 温暖的琥珀色调、柔和对比度、发光般的高光 |
| 北欧极简 | 冷调、低饱和度的蓝色与灰色、低对比度、画面简洁 |
常见问题
- 指令含糊不清,例如“让它看起来很酷”,会导致结果不一致。请具体描述想要的效果。
- 忽略时间背景,而编辑本应应用于片段中的特定时刻。
- 期待精准到帧的剪切,但只用自然语言描述。要精确剪辑,请注明时间戳。
- 过度处理,一次请求太多特效。分阶段应用编辑,效果最佳。
- 忘记考虑源素材质量。低分辨率输入会限制画质增强的效果。
相关技能
文档
评论
支持 Markdown