AI-based API Aware Image and Video Generation System

This project provides a workflow for generating AI-driven, API-aware images and videos based on text prompts, storing the generated content and logs in databases, and serving the results through a web interface.

---

Overview

1. Text-to-Image and Text-to-Video Generation

Using OpenAI’s DALL·E 3 and RunwayML’s video generation model, this system takes a user prompt and generates 5 images and 5 corresponding videos.

• The image_video_generation.py code uses a user-provided text prompt to generate images via the DALL·E 3 image generator.
• Each generated image is saved in generated_content/<user_id> as image_1.png, and so on.
• Each image is then converted to Base64 format and, along with the text prompt, sent to the RunwayML API to generate a 5-second video. This video is saved in the same directory as the corresponding image (e.g., video_1.mp4).

Note: More images and videos can be generated by changing the n parameter in the generate_images_and_videos() function. However, this may incur significant costs.

generate_images_and_videos(user_id, prompt, n=5, image_size="1024x1024", video_duration=5)

• Images: Stored in generated_content/<user_id>/image_x.png.
• Videos: Stored in generated_content/<user_id>/video_x.mp4.

2. Storing and Managing Content

The generated content details (prompts, file paths, status, timestamps) are stored in a database (ai_generation.db). Once generation is complete, the status is updated to "Completed."

3. User Access and Web Page Display

A Flask web application (web.py) allows users to access their generated images and videos by providing a user_id.

• If the content is still processing, a "Processing" page is displayed.
• Once completed, a gallery of images and playable videos is shown.

User actions (login attempts and content views) are logged in a separate database (user_logs.db).

4. Notifications

When content generation is complete, the system prints a notification message to the terminal, including a link to the web page where the user can view their content.

Directory Structure

• generated_content/: Holds generated images and videos for each user.
• templates/: Contains HTML templates for the gallery and processing pages.
• main.py: CLI script to trigger generation and send notification once done.
• web.py: Flask web application serving the generated content and logging user actions.
• database.py: Manages the ai_generation.db for storing user content and metadata.
• config.py: Loads environment variables and configuration.
• image_video_generation.py: Contains functions for generating images and videos.
• notifications.py: Handles user notifications (in-terminal message).
• requirements.txt: Lists Python dependencies.
• .env: Contains API keys and other secrets (not checked into version control).

Prerequisites

• Python 3.11.2
• A .env file with the following variables:

  OPENAI_API_KEY=<your_openai_api_key>
  RUNWAYML_API_SECRET=<your_runwayml_api_key>

• Required Python packages

  pip install -r requirements.txt

• Export RunwayML API key (for Linux/Mac; for Windows, set it via .env):

  export RUNWAYML_API_SECRET=<your_runwayml_api_key>

Usage

1. Content Generation Run main.py to start the generation process:

python main.py <user_id> "<prompt>"

Example:

python main.py user123 "A serene landscape with mountains and a river at sunset."

This will:

• Insert a new record into ai_generation.db with status "Processing."
• Generate 5 images and 5 videos using the prompt.
• Update the database record to "Completed" once done.
• Print a terminal notification indicating that content is ready.

2. Serving the Content Start the Flask server:

python web.py

Access the content in your browser:

http://localhost:5000/?user_id=user123

If still processing, you’ll see a "Processing" page. Once completed, a gallery of images and videos will be displayed.

3. Logs

The user_logs.db database logs user interactions:

"Login" when a user visits / with user_id.

"View Content" when a user views an image or video file.

Notes

• Ensure that generated_content/ directory is writable.
• The code currently uses eval() to parse lists of file paths stored as text in the database. For a production environment, prefer json.loads() for safer parsing.
• RunwayML and OpenAI generation may take some time. The processing page remains active until the main.py script completes generation.
• The notification is currently in-terminal only. You can extend notifications.py for email or other services if desired.