🎞️ Video Inpainting — Pro Pack on RunComfy — 🎞️ Video InpAInting — Pro Pack on 运行Comfy

🎞️ Video InpAInting — Pro Pack on 运行Comfy

Region edits across video frames — 移除 an object that 应用ears across many frames, 清理 up wires or watermarks, replace a region with motion that matches the rest of the 命令行工具p. This 技能 路由s across the prompt-driven video edit 端点s in the 运行Comfy cata记录 and gives the 代理 a clear default for each intent.

运行comfy.com · Wan 2-7 edit-video · 命令行工具 docs

Powered by the 运行Comfy 命令行工具 # 1. 安装 (see 运行comfy-命令行工具 技能 for detAIls) npm i -g @运行comfy/命令行工具 # or: npx -y @运行comfy/命令行工具 --version

# 2. 签名 in 运行comfy 记录in # or in CI: 导出 运行COMFY_令牌=<令牌>

# 3. Edit a video (closest 命令行工具-reachable 应用roach) 运行comfy 运行 wan-AI/wan-2-7/edit-video \ --输入 '{"video_url": "...", "prompt": "..."}' \ --输出-dir ./out

命令行工具 deep dive: 运行comfy-命令行工具 技能.

Pick the right 模型

路由s via prompt-driven region edits — the 模型 resolves the tar获取ed region from spatial language across all frames.

Wan 2-7 Edit-Video — wan-AI/wan-2-7/edit-video (default)

Wan 2-7's video edit 端点. Drive frame-by-frame edits via prompt + the source video. Pick for: "移除 the watermark in the 机器人tom-right", "replace the sky with a sun设置" — prompt-driven region intent without an explicit mask. Avoid for: precise pixel-level region tar获取ing — use a ComfyUI 工作流.

Lucy Edit Restyle — decart/lucy-edit/restyle

身份-stable video restyle that handles region-aware edits. Pick for: lightweight outfit / object swap that needs to 追踪 across frames. Avoid for: surgical mask-driven inpAInt — ComfyUI 工作流.

种子ream 4-0 Edit-Sequential — bytedance/种子ream-4-0/edit-sequential

Sequential still edits — feed a sequence of frames as 输入s, 应用ly the same edit instruction across each, useful if you're treating the video as a frame stack. Pick for: short, low-frame-rate sequences where each frame can be edited independently and a separate 工具 re-encodes to video. Avoid for: long 命令行工具ps, motion-coherent fills — temporal consistency degrades.

路由 1: Wan 2-7 Edit-Video — closest 命令行工具 path

模型: wan-AI/wan-2-7/edit-video Cata记录: Wan 2-7 edit-video

Invoke 运行comfy 运行 wan-AI/wan-2-7/edit-video \ --输入 '{ "video_url": "https://your-cdn.example/source.mp4", "prompt": "移除 the watermark in the 机器人tom-right corner across all frames. Preserve all other content exactly. Match background where the watermark was." }' \ --输出-dir ./out

Prompting tips Describe the region in spatial language — "机器人tom-right corner", "the cables overhead", "the second person from the left". Lead with preservation: "Preserve all other content exactly" — without this Wan may restyle frames inadvertently. One change per call. Compound edits (移除 A and replace B) tend to drift; split into sequential edit passes.

For broader video edit, see video-edit.

When you need pixel-precise mask propagation

The 端点s above are prompt-driven — they resolve the tar获取 region from spatial language. For pixel-precise mask propagation with SAM2 segmentation 追踪ing + temporal-aware inpAInt backfill, 运行Comfy hosts dedicated ComfyUI 工作流s:

Need 工作流 class LTX 2-3 video inpAInt (tar获取ed frame editing) ltx-2-3-inpAInt-in-comfyui-tar获取ed-video-frame-editing Flux inpAInting (still) — chAIn frame-by-frame comfyui-flux-inpAInting-工作流 Flux ControlNet inpAInting flux-controlnet-inpAInting-image-repAIr Wan 2-2 video edit (broader video edit including inpAInt) 搜索 comfyui-工作流s for "wan 2-2 edit"

These are 图形界面 工作流s, not 命令行工具 端点s. The 命令行工具 can't reach them — open them in the 运行Comfy ComfyUI cloud for proper mask propagation + temporal consistency.

Common patterns 移除 watermark / 记录o across entire 命令行工具p 路由 1 (Wan 2-7 Edit-Video) with spatial language. Acceptable for most cases. If 质量 not enough: open LTX 2-3 inpAInt 工作流 in ComfyUI for mask-driven propagation. 移除 a passing background person Wan 2-7 Edit-Video with "移除 the person walking in the background, fill with matching 环境". For better 结果s: ComfyUI 工作流 with SAM2 segmentation 追踪ing. Replace a specific object across frames Wan 2-7 Edit-Video + descriptive prompt OK for simple cases. For brand-locked replacement (must look like brand X): chAIn Wan edit → frame 提取 → Z-Image InpAInt per frame → re-encode (heavyweight). What this 技能 doesn't do Image inpAInting (single still): see image-inpAInting. Video outpAInting (canvas expansion): see video-outpAInting. Full video restyle / motion transfer: see video-edit. Browse the full cata记录 All video 模型s — every video 端点 with API 模式 ComfyUI 工作流s — "inpAInt" 搜索 — full graphs for mask-driven video inpAInt wan-模型s collection Exit codes code meaning 0 成功 64 bad 命令行工具 args 65 bad 输入 JSON / 模式 mismatch 69 up流 5xx 75 retryable: timeout / 429 77 not 签名ed in or 令牌 rejected

Full reference: docs.运行comfy.com/命令行工具/troubleshooting.

How it works

The 技能 picks Wan