In short
- The app runs on your computer and your projects stay there. You don't need an account to use it.
- If you sign in, we keep your name, profile picture and what you use changji Cloud for — enough to run the service and show you your usage.
- The app sends anonymous usage statistics — never your prompts, stories, files or keys. You can turn it off in Settings.
- We don't sell your data, show ads or use third-party analytics, and we don't train models on your content.
What we collect
| What | Why |
|---|---|
| Account: your Google account ID, name and profile picture. We request your email address from Google but only use it as your display name if your profile has no name. | To sign you in and show who's signed in. |
| Devices: the name and system of each device you authorize, when it was signed in and last used. Device keys are stored only as a hash. | So you can see and revoke them. |
| Usage: LLM tokens by model, GPU time, length of rendered video, number of jobs, by day. | Quotas, plans and your usage page. |
| Cloud requests: the text you send to cloud LLMs and the images you send for rendering. | Passed on to run the job. We don't keep LLM requests. Rendering inputs and results are cached on our GPU servers for a limited time so the app can fetch them. |
| Visit logs: for each request to this site — time, IP address, approximate location derived from it, browser and system, referring page, language, the page requested (with secrets in the address removed), status and size, and the account or device if signed in. | Security, abuse prevention and understanding how the site is used. Cookies and credentials are never logged. |
| App usage statistics: anonymous, sent by the changji app — see below for exactly what. | To see how many people use changji, on what hardware, and which steps fail, so we can fix them. |
| Orders: plan, amount, currency, time and payment status. | To give you what you paid for and keep accounts. Card details are handled by Stripe and never reach us. |
App usage statistics
The changji app — the desktop app and the command-line engine — sends anonymous usage statistics to this site. It is on by default; the app says so the first time it starts, and you can turn it off at any time in Settings → General → Send anonymous usage statistics, or by setting the environment variable CHANGJI_TELEMETRY=0 or DO_NOT_TRACK=1. The same settings page shows exactly what would be sent.
| What is sent | When |
|---|---|
| A random install ID generated on your computer (not linked to your account or your computer's name); the version, desktop app or engine, system, architecture, graphics build and interface language. | With everything below. |
| Whether the app is idle, writing or rendering. | Every 5 minutes while the app is open, and when rendering starts or stops. |
| Events: installed, a chapter written, a shot rendered, a film joined (with its length in minutes). | When they happen. |
| Daily counts: chapters, shots, voice lines and finished minutes; how each step went (succeeded, failed, cancelled, time taken); which quality checks sent work back. Graphics card model, graphics memory and RAM; whether the language model is local, a remote service or a command — for a remote service only which provider, never its address or key; which video, image and voice models are selected. | Once a day. |
Never sent: prompts, story or script text, project or file names, images, video or audio, API keys, server addresses, your user name or your computer's name.
Your IP address is used only to look up the city when the statistics arrive (coordinates rounded to about 10 km) and is not stored. Usage statistics are kept as long as visit logs.
Cookies
We use two cookies: cj_sess keeps you signed in, and cj_lang remembers your language. There are no advertising or tracking cookies.
Who we share with
- Google — to sign you in.
- Cloudflare — delivers this site; it sees your IP address and may provide an approximate location.
- ip-api.com — when Cloudflare doesn't provide a location, we look up the IP address there.
- Upstream model providers — cloud LLM requests are forwarded to the provider that runs the model, under their own terms.
- Stripe — processes payments.
- GitHub — hosts the download files.
We may also disclose information if the law requires it.
Retention
Account, device and order records are kept while your account exists. Usage is kept per month. Visit logs and app usage statistics are deleted after 365 days by default. When you delete your account we delete your account and device records; records we must keep by law, such as payment records, are kept only as long as required.
Your choices
- Use the app without an account — nothing about your projects reaches us.
- Turn off app usage statistics in Settings → General, or with
CHANGJI_TELEMETRY=0orDO_NOT_TRACK=1. - Revoke devices at any time on My account.
- Ask us for a copy of your data, to correct it or to delete your account: contact us.
The service isn't directed at children under 13, and we don't knowingly collect their data.
Security
Sessions and device keys are stored only as hashes, the site is served over HTTPS, and administrative access is limited. No system is perfectly secure; if we learn of a breach affecting you, we'll let you know.
We'll update this policy when our practices change and change the date above. Questions? Contact us.
概要
- 程序运行在你的电脑上，片子也保存在你的电脑上。不登录账号也能使用。
- 如果你登录，我们会保存你的名字、头像，以及你使用场记云的情况——仅够运行服务、向你展示用量。
- 程序会发送匿名的使用统计，从不包含你的提示词、故事、文件和密钥；可以在设置中关闭。
- 我们不出售你的数据，不投放广告，不使用第三方统计分析，也不用你的内容训练模型。
我们收集什么
| 内容 | 用途 |
|---|---|
| 账号：你的 Google 账号 ID、名字和头像。我们会向 Google 请求你的邮箱地址，但仅在你的资料没有名字时把它用作显示名。 | 登录，并显示当前登录的是谁。 |
| 设备：你授权的每台设备的名称和系统、登录时间和最近使用时间。设备钥匙只保存其哈希值。 | 让你查看和吊销设备。 |
| 用量：按模型统计的大模型 token、显卡时间、生成视频的时长、任务数量，按天记录。 | 额度、方案和用量页面。 |
| 云端请求：你发送给云端大模型的文字，以及为出片发送的图片。 | 转交执行任务。我们不保存大模型请求；出片的输入和结果会在我们的显卡服务器上缓存一段时间，供程序取回。 |
| 访问日志：对本网站的每个请求：时间、IP 地址及据此推断的大致位置、浏览器和系统、来源页面、语言、请求的页面（地址中的密钥会被去除）、状态和大小，登录时还包括账号或设备。 | 安全、防止滥用，以及了解网站的使用情况。Cookie 和凭据从不记录。 |
| 软件使用统计：匿名，由场记程序发送，具体内容见下文。 | 了解有多少人在用场记、用的是什么硬件、哪些步骤会失败，以便改进。 |
| 订单：方案、金额、币种、时间和付款状态。 | 为你提供所购买的服务并留存账目。卡号信息由 Stripe 处理，不会到达我们这里。 |
软件使用统计
场记程序（桌面端和命令行引擎）会向本网站发送匿名的使用统计。默认开启；程序首次启动时会告知你，你可以随时关闭：设置 → 通用 → 发送匿名使用统计，或者设置环境变量 CHANGJI_TELEMETRY=0 或 DO_NOT_TRACK=1。同一个设置页上可以查看会发送的全部内容。
| 发送的内容 | 何时发送 |
|---|---|
| 在你的电脑上随机生成的安装编号（与你的账号、电脑名称都不关联）；版本、桌面端还是引擎、系统、架构、显卡版本和界面语言。 | 随下面每一项一起发送。 |
| 程序当前是空闲、写作中还是出片中。 | 程序开着时每 5 分钟一次，开始或结束出片时各一次。 |
| 事件：新安装、写完一章、出完一镜、接成一部片子（附成片分钟数）。 | 发生时。 |
| 按天的计数：章数、镜数、配音句数和成片分钟；每一步的结果（成功、失败、取消、耗时）；哪些质量检查把结果退回。显卡型号、显存和内存；大模型用的是本地模型、远程服务还是命令行——远程服务只记是哪一家，不记地址和密钥；选用的出片、出图和配音模型。 | 每天一次。 |
从不发送：提示词、故事和剧本的文字、片子和文件的名称、图片、视频和音频、API Key、服务器地址、你的用户名和电脑名称。
你的 IP 地址只在统计到达时用于查询所在城市（经纬度取整到约 10 公里），不会保存。使用统计的保存期限与访问日志相同。
Cookie
我们使用两个 Cookie：cj_sess 用于保持登录，cj_lang 用于记住你选择的语言。没有广告或追踪 Cookie。
与谁共享
- Google：用于登录。
- Cloudflare：负责传输本网站，会看到你的 IP 地址，并可能提供大致位置。
- ip-api.com：Cloudflare 未提供位置时，我们在此查询 IP 地址所在位置。
- 上游模型服务商：云端大模型请求会转交给运行该模型的服务商，适用其自身条款。
- Stripe：处理付款。
- GitHub：托管下载文件。
在法律要求时，我们也可能披露信息。
保存期限
账号、设备和订单记录在你的账号存续期间保存。用量按月保存。访问日志和软件使用统计默认 365 天后删除。你删除账号后，我们会删除你的账号和设备记录；法律要求保留的记录（如付款记录）仅保存到规定期限为止。
你的权利
- 不登录也可以使用程序——你的片子不会有任何内容到达我们这里。
- 在设置 → 通用中关闭软件使用统计，或者设置
CHANGJI_TELEMETRY=0或DO_NOT_TRACK=1。 - 随时在我的账号中吊销设备。
- 如需获取你的数据副本、更正数据或删除账号，请联系我们。
本服务不面向 13 岁以下的儿童，我们不会有意收集他们的数据。
安全
会话和设备钥匙只保存哈希值，网站通过 HTTPS 提供，管理权限受到限制。没有绝对安全的系统；如果发生影响到你的数据泄露，我们会通知你。
我们的做法发生变化时会更新本政策并修改上方日期。如有疑问，请联系我们。