中文 English Technical

牧云图片 — 隐私政策

最后更新 / Last updated: 2026-08-11

中文版

牧云图片不收集、不上传、不分享你的任何数据。 你的图片、编辑记录、偏好设置全部只在你这台 Mac 本地处理和保存。 没有后端服务器,没有云同步,不内嵌任何第三方分析、追踪或广告 SDK。 所有图像处理——滤镜、调色、标注、马赛克、内容感知消除与 AI 消除——全部在你的设备上完成。

我们处理什么,在哪处理

数据用途处理/存储位置
你打开的图片显示与编辑仅内存,经 macOS 沙盒"用户选择文件"权限读取,保存时写回你指定位置
编辑内容(标注/滤镜/调整/消除区域)当前编辑会话仅内存,随导出图片保存,不单独持久化
AI 消除模型(LaMa,约 200MB)本地 AI 消除推理设备本地 Application Support/MuyunImage/Models/LaMa,可随时删除
打印任务你主动打印图片macOS 系统打印服务,App 不留存副本

我们 不 做的事

网络通信

本 App 仅在一种情况下发起网络连接,且完全由你主动触发:

  1. 下载 AI 消除模型 — 当你在"消除"工具中选择"AI 智能消除"并点击下载时,App 通过 HTTPS 从 GitHub Releases(开源项目 CoreMLaMa 的公开构建)下载 LaMa CoreML 模型;直连失败时会尝试公共 GitHub 镜像加速站。请求中不包含你的任何图片或个人信息。

除此之外,App 不与任何服务器通信。若不使用 AI 消除(内容感知消除无需模型),App 可在完全断网环境下使用全部功能。

文件访问

牧云图片运行在 macOS App Sandbox 中,仅能访问你通过打开/保存对话框或拖拽主动授予的文件,无法扫描或读取磁盘上的其他内容。

数据删除

English Version

MuyunImage collects no data, transmits nothing to remote services, and shares nothing with third parties. Your images, edits, and preferences are processed and stored exclusively on your own Mac. There is no backend server, no cloud sync, and no embedded analytics, tracking, or advertising SDK. All image processing — filters, color adjustment, annotations, mosaic, content-aware removal, and AI removal — happens entirely on your device.

What we process and where

DataPurposeLocation
Images you openDisplay & editingIn-memory only, via macOS sandbox user-selected-file permission; saved back where you choose
Edits (annotations / filters / adjustments / masks)Current sessionIn-memory only, baked into exported images; never persisted separately
AI removal model (LaMa, ~200 MB)On-device AI inpaintingDevice-local Application Support/MuyunImage/Models/LaMa, removable anytime
Print jobsPrinting you initiatemacOS print system; the app keeps no copy

What we do not do

Network traffic

The app initiates network traffic in exactly one case, explicitly triggered by you:

  1. AI model download — Choosing "AI Removal" and tapping download fetches the LaMa CoreML model over HTTPS from GitHub Releases (public build of the open-source CoreMLaMa project), with public GitHub mirror CDNs as fallback. The request contains none of your images or personal information.

Outside of that, the app talks to no servers at all. Without AI removal (the content-aware engine needs no model), the app works completely offline.

File access

MuyunImage runs inside the macOS App Sandbox and can only access files you explicitly grant via open/save dialogs or drag & drop.

Data deletion

Technical Statement

For App Store reviewers and security-conscious users: