视频|Kling Motion Control 动作控制
Transplant an actor's moves onto your character image
- 角色图像
- 📁_本地保存路径
- 🆔_任务ID
- 🔗_视频云端直链
- 📄_完整日志
- 🎬_视频输出
This is the pack's most distinctive Kling node, and the most useful for anyone doing character/IP work: it takes a reference video's motion and transplants it onto your own character image. You feed it a static 角色图像 (the character who should move) plus a 动作参考视频URL (the moves to copy), and it returns a video of your character performing those moves while keeping the character's identity - same face, same outfit, same look. This is the "AI digital human" and "TikTok remix" trick, and it's the kind of thing Kling does better than almost anyone.
The mechanism is worth understanding because it's not magic: this is motion transfer, not full re-generation. The reference video supplies the motion; the character image supplies the identity. It works best with full-body motion on a clean background - a cluttered action video gives the motion extractor noise to chew on. And it inherits every limit of the underlying Kling v2.6/v3 model, which the node exposes through the 模型版本与模式 dropdown (std 720P vs pro 1080P).
How it works
The reference video needs to be a public URL (动作参考视频URL); for a local file there's 本地动作视频路径, which the node uploads to Tikpan first - for big videos, a URL beats an upload every time. The 动作描述 field is a free-text supplement ("keep the expression natural, full-body movement") that refines the transfer. 视频时长 defaults to auto, following the reference video's length, and 画面比例 defaults to auto too - both make sense when the source video defines the output.
Outputs are the standard five: 🎬_视频输出 (VIDEO), 📁_本地保存路径, 🆔_任务ID, 🔗_视频云端直链, 📄_完整日志.
The inputs that matter
角色图像- your character, static, ideally full-body and well-lit.动作参考视频URL- the moves to copy; clean, single-subject footage transfers best.模型版本与模式-kling-v2-6|std|720Pby default; pro for 1080P.动作描述- nudge the transfer: keep the face natural, avoid artifacts, etc.本地动作视频路径- only if your source is local and under the upload cap.
Install and first run
cd ComfyUI/custom_nodes
git clone https://github.com/htrert/ComfyUI-Tikpan-Pro
Restart (or Manager → "Tikpan"), paste your sk- key, load the character image, paste the reference URL, run at the default settings. Start with a short, single-action reference clip - motion transfer is easier to judge and cheaper to iterate on when the source motion is one clean action rather than ten seconds of everything.
Common issues
The pack-wide staples (401/402/429, timeouts - raise 最长等待秒数 for pro mode or long references - and the charge-with-no-file case, always check 完整日志). Motion-control-specific: the reference video being too busy is the top quality killer (motion extraction degrades on cuts and cluttered scenes), and a character image that isn't full-body will fight the reference motion. Local video uploads are the flakiest path - put the file on a public URL instead if you can.
Inputs (15)
| Name | Type | Default | Description |
|---|---|---|---|
| 💰_福利_💰 | COMBO | 1 options: 🔥 0.6元RMB兑1美元余额 | 全网底价 👉 https://tikpan.com | |
| 获取密钥请访问 | COMBO | 1 options: 👉 https://tikpan.com (官方授权Key获取点) | |
| API_密钥 | STRING | sk- | Tikpan 平台的 API 密钥,以 sk- 开头,从 https://tikpan.com 获取 |
| 角色图像 | IMAGE | 目标角色的静态图像,模型会让这个角色做参考视频里的动作 | |
| 动作参考视频URL | STRING | 公开可访问的视频 URL。也可以在下面填本地视频路径,二选一。 | |
| 动作描述 | STRING | 保持角色身份一致,将参考视频中的动作精准迁移到角色图像,生成自然流畅的全身动作视频。 | 对动作迁移效果的补充描述,例如『保持人物表情自然』 |
| 模型版本与模式 | COMBO | kling-v2-6|std|720P | 选择 Kling 版本和质量模式:std 标准 720P;pro 高质 1080P |
| 视频时长 | COMBO | 自动按参考视频|auto | 生成视频秒数;auto=跟随参考视频时长 |
| 画面比例 | COMBO | 自动|auto | 视频比例:auto 跟随参考;常用 16:9 横屏、9:16 竖屏 |
| 本地动作视频路径opt | STRING | 填写本地 mp4/mov/webm 路径后,节点会先上传到 Tikpan,再提交动作控制任务。 | |
| 中转站地址opt | COMBO | https://tikpan.com | Tikpan 中转站地址,一般保持默认即可 |
| 最长等待秒数opt | INT | 90060–3600 | 等待视频生成完成的最长秒数;长视频/pro 模式建议加大 |
| 查询间隔秒数opt | INT | 85–60 | 轮询任务状态的间隔秒数 |
| 校验HTTPS证书opt | BOOLEAN | false | 默认关闭以兼容部分网络;遇到 SSL 问题可保持关闭 |
| 跳过错误opt | BOOLEAN | false | 开启后异常时返回空,不打断后续工作流 |
Outputs (5)
| Name | Type | Description |
|---|---|---|
| 📁_本地保存路径 | STRING | — |
| 🆔_任务ID | STRING | — |
| 🔗_视频云端直链 | STRING | — |
| 📄_完整日志 | STRING | — |
| 🎬_视频输出 | VIDEO | — |