AIVeo3 MVP Demo

中文版

Introduction

This is a simple MVP demo about connecting orginal API from google from worldwide but still follow it's policy. If you want to try to generate videos with Google Veo3, this project should be your best practice to start.

It's...the best i can tell...u need to try the details about it and...keep your mouth shut...don't brag

For this version(there will be no futher update😛😜😝), it can do:

1. text to video
2. image and text to video
3. no more

How to use

Andthe bugs in this version:

1. The generate order may not sync the real status after you submit it, you will need to refreash the page to see it. And also may not sync the result...so you need to....
2. The time generation cost, is not showing real time I'm happy anyone can solve this, I can't solve it... This is a tricky part, it's only fit for Linux, and if you don't know anything about it......just follow the leads(or you can just ask Gemini🤣🤣🤣)

First

It's the most important step, you need to prepare two server(if you are in State...forget this part), and at least of it need to be in the US, it's for the backend program.

Second

Download o Git clone codes to your server

Third

Deal with the backend first. Use command tools get into the backend's dir, then run the command below

pip install -r requirements.txt

The requirements.txt is generated automatically, if it pop out sng warning, don't worry about it(only if tells you already have dependencies o your network connect is failed), just keep going, we can test it later.

Forth

Get your api key from google cloud(u can learn form google), then paste it to the first key GOOGLE_API_KEY in the config.json, the file will let u know where you need to paste it.

Fifth

Run the app.py in the command line，if it goes well, then goto next step. If it goes wrong...check the "Third" step, if it's still not work, as the Gemini please...🙊🙈🙈

Sixth

Make sure your Linux or your server is opened and listening on port 5000(you can change it in app.py file)

Seventh

Keep the command line in step "Fifth" running, then deploy the frontend. I use the openresty(some sub version of Nginx), the key is to write a reverse proxy to your backend(which means the app.py we talk in step "Fifth"). Here is my whole config in the openresty(include the reverse proxy), and i deploy it with 1panel(highly recommand). You will need to replace these config wo your own :server_name, access_log, error_log, location /api/, proxy_pass, root, ssl_trusted_certificate.

server {
    listen 80 ; 
    listen 30030 ; 
    listen 30300 ; 
    server_name replace me; 
    index index.php index.html index.htm default.php default.htm default.html; 
    proxy_set_header Host $host; 
    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; 
    proxy_set_header X-Forwarded-Host $server_name; 
    proxy_set_header X-Real-IP $remote_addr; 
    proxy_http_version 1.1; 
    proxy_set_header Upgrade $http_upgrade; 
    proxy_set_header Connection "upgrade"; 
    access_log /www/sites/veo3/log/access.log; 
    error_log /www/sites/veo3/log/error.log; 
    location ^~ /.well-known/acme-challenge {
        allow all; 
        root /usr/share/nginx/html; 
    }
    # 后端API反向代理
    location /api/ {
        proxy_pass http://your_server_ip_or_domain:5000/; 
        proxy_set_header Host $host; 
        proxy_set_header X-Real-IP $remote_addr; 
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; 
        proxy_set_header X-Forwarded-Proto $scheme; 
    }
    add_header 'Access-Control-Allow-Origin' 'http://your_server_ip_or_domain:5000' always; 
    add_header 'Access-Control-Allow-Methods' 'GET, POST, OPTIONS'; 
    root /www/sites/veo3/index; 
    ssl_trusted_certificate /www/sites/veo3/ssl/fullchain.pem; 
}

If you use 1panel like me, all you need to do paste and alternate is the # 后端API反向代理 part

My site is not support HTTPs, so you need to figure it your own if you want.

Eighth

Start the site, see if it works. My package has few test and demo history file, if their shows up, then try to generate a video. After you generate a video, you should know this project runs successful. Then you can goto next step, otherwise...copy the oder id(they looks like this 0d084f96-3540-49e6-820c-ec9a56b804c6) and check the log in the backend's dir /log, you will finout what happends.

Ninth

Kill the command line process we build in the step "Fifth". Alternate the system service in the system service folder, you need to change these to your real configs: WorkingDirectory, ExecStart. Then put the file to this directory /etc/systemd/system/ (my system is Debian, so it goes here, u may need to google your own OS about it).

Tenth

Run commands below one by one:

systemctl daemon-reload

systemctl start video-generator.service

systemctl status video-generator.service

By now, you can see a bunch of things, there will be some words show up green and before it have some words like "Active: ", then you know it works.

Other thing...

• The service part is not have to be done, just can make it run all the time even after your machine reboot.
• If you find the file have some problem and seems nthing wrong, youcan try use tool like "dos2unix" convert all file to unix, because when i committ this on PC...so...