Actions24
- Audio Actions
- Image Actions
- Video Actions
- Animate Mix | 视频换人
- Animate Move | 图生动作
- Emoji Video | Emoji 视频生成
- Image to Video (First & Last Frames) | 图生视频(首尾帧)
- Image to Video (First Frame) | 图生视频(首帧)
- Video Continuation | 视频续写
- Reference to Video | 参考生视频(wan2.6)
- Text to Video | 文生视频
- Video Editing & Effects | 视频编辑与特效
- Video Retalk (Lip Sync) | VideoRetalk 口型复刻
- Task Actions
- Vision Actions
- File Actions
Video → Video Editing & Effects | 视频编辑与特效
AI-generatedSummary
Perform advanced video editing and effects using Aliyun Bailian's AI-powered editing models, allowing tasks such as image-reference based generation, mask-based video edits, seamless clip extensions, outpainting for canvas expansion, and video repainting with various conditioning signals.
Inputs
- Model (required) — Choose the unified video editing model (Wan 2.7 Video Edit or Wanx2.1 VACE Plus) to define available editing functions and behavior.
- Function (required) — Select the specific editing function based on the chosen model, e.g., image reference generation, mask-based edits, video extension, outpainting, or repainting.
- Prompt (required) — Provide a scene description prompt to guide the editing or generation process.
- Negative Prompt — Optionally specify a negative prompt to exclude undesired elements (available with Wan 2.7 Video Edit model).
- Reference Images — List URLs or data URIs of reference images to influence video generation (up to 4 for Wan 2.7 model).
- Video URL — URL of the source video clip to be edited or stylized.
- Resolution | 分辨率档位 — Set video output resolution tier (1080P or 720P) for Wan 2.7 Video Edit model.
- Ratio | 宽高比 — Aspect ratio of the output video, or empty to follow the input video's ratio.
- Duration | 截断时长 — Truncate the input video duration from the start; 0 to keep original length.
- Audio Setting — Configure audio regeneration: auto (model decides) or origin (keep source audio) for Wan 2.7 model.
- Auto Enhance Prompt — Enable or disable automatic prompt expansion to enhance editing quality for Wan 2.7 model.
- Advanced Options | 高级选项 — Additional settings including embedding AI watermark, specifying deterministic seed, and controlling seed randomization (Wan 2.7 model).
- Mask Image URL — Provide a PNG mask image URL to define the edit region for mask-based video edits (video_edit function).
- Mask Frame Index — Frame index for the mask image upload; defaults to the first frame (0).
- Mask Tracking Mode — Select how the mask region is tracked across video frames: tracking or static.
- Mask Expand Ratio — Set padding ratio around the mask area to stabilize edges during edits.
- First Clip URL — Specify a video clip URL to extend seamlessly (video_extension function).
- Control Condition — Choose conditioning signal (depth, pose, canny, sketch) for video repainting.
- Targets (Obj/Bg) — Select layers (object and/or background) to apply edits when using layered templates.
- Outpainting Top Scale — Top edge canvas expansion scale factor for video outpainting.
- Outpainting Bottom Scale — Bottom edge canvas expansion scale factor for video outpainting.
- Outpainting Left Scale — Left edge canvas expansion scale factor for video outpainting.
- Outpainting Right Scale — Right edge canvas expansion scale factor for video outpainting.
- Wait for Task Completion — Choose whether to synchronously wait for the video editing task to complete or to receive the task ID immediately.
- Poll Interval (MS) — Interval in milliseconds between polling attempts when waiting for task completion.
- Max Wait (Seconds) — Maximum duration in seconds to poll before timing out when waiting for task completion.
Output shape
a structured JSON object containing task status and result URLs after video editing or the immediate task ID if waiting is disabled
If 'Wait for Task Completion' is enabled, the node polls the remote task endpoint until success or failure, returning the completed result or error; otherwise, it returns the task ID immediately for asynchronous tracking.
Examples
Example 1: Apply an image reference based video edit using Wanx2.1 VACE Plus model
Set Model to 'Wanx2.1 VACE Plus', Function to 'Image Reference', provide 'Prompt' describing the desired scene, list Reference Images, and specify the source 'Video URL' if editing an existing clip.
Example 2: Perform mask-based localized edits on a video using Wanx2.1 VACE Plus
Choose Model 'Wanx2.1 VACE Plus', Function 'Video Edit (Mask Based)', upload a mask PNG via 'Mask Image URL', set the mask frame index and tracking mode, and provide the edit 'Prompt'.