https://github.com/postor/ai-art-picture-diary

an app that automatically generate pictures for your diary, with help of ChatGPT and Stable Diffusion
https://github.com/postor/ai-art-picture-diary

Last synced: 2 months ago
JSON representation

an app that automatically generate pictures for your diary, with help of ChatGPT and Stable Diffusion

• Host: GitHub
• URL: https://github.com/postor/ai-art-picture-diary
• Owner: postor
• License: mit
• Created: 2023-04-26T13:27:29.000Z (about 2 years ago)
• Default Branch: main
• Last Pushed: 2023-05-10T07:00:15.000Z (about 2 years ago)
• Last Synced: 2025-01-25T18:10:26.508Z (4 months ago)
• Language: Java
• Size: 11.2 MB
• Stars: 2
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# ai-art-picture-diary

这是一个使用 ChatGPT 和 Stable Diffusion 自动给你的日记配图的应用

an app that automatically generate pictures for your diary, with help of ChatGPT and Stable Diffusion

[![screenshot](https://img.youtube.com/vi/oh4CcihBrHg/0.jpg)](https://www.youtube.com/watch?v=oh4CcihBrHg)

## 项目由来 | why this project

- 总体来讲这是一个调研，学习如何使用 Stable Diffusion, 以及了解它能做到什么程度 | it's mostly a research, to learn how to use Stable Diffusion and how good can it be

- 这个日记配图的想法也是因为我刚好有个该写日记了的女儿 | this idea  come to me thanks to my 6 year old daughter

## 项目启动 | set up

- 启动 nats, 他负责传递 nodejs-server 和 stable-diffusion-worker 之间的消息, 运行 `cd nats && docker compose up` 或者启动你自己的 nats 但需要记得调整消息大小限制，默认只有 1M 是不行的

- 启动 stable diffusion worker, 他负责从 prompt 生成图片, 提前安装 python3 和 pip 包 `diffusers`,`torch`,`asyncio`,`nats` 然后运行 `cd stable-diffusion-worker && python3 worker.py`

- 启动 nodejs server, 它接收客户端的请求并响应, 提前安装 NodeJS 并设置环境变量 `OPENAI_API_KEY`, 运行 `cd nodejs-server && npm i && npm run dev`

- 构建并运行 android app, 使用 android studio 打开, 修改服务端 IP (搜索 `10.0.0.197` 并替换为实际 IP), 接入手机并开启 adb 调试 然后选择在手机启动 `app`

- bring up nats, it's for messaging between nodejs-server and stable-diffusion-worker, `cd nats && docker compose up` or you can use your own nats and make sure allow large messages

- bring up stable diffusion worker, it provides picture from prompt, make sure you have python3 and pip packages `diffusers`,`torch`,`asyncio`,`nats` then `cd stable-diffusion-worker && python3 worker.py`

- bring up nodejs server, it receive client requests and response, make sure you have NodeJS and set env `OPENAI_API_KEY`, then `cd nodejs-server && npm i && npm run dev`

- build and run android app, open with android studio, change server IP (search `10.0.0.197` and replace your server IP), then plugin your device with adb debug and run `app` on your device