How pr0tel/chooseone works
Main Components
• Video Generation Engine (main.py): This is the entry point controlling the workflow.
• AI Services Integration:
– GPT: Used to generate questions and text prompts.
– Kandinsky: Used to transform text prompts into images.
• Media Assets:
– Assets Folder: Contains pre-made media like audio (audio.mp3) and video backgrounds (first_bg.mp4) used during video compilation.
• External Dependencies:
– Python libraries specified in requirements.txt.
• Output:
– The final compiled video is produced and stored in the root folder.
Relationships and Data Flow
• The engine (main.py) initiates the process by calling the GPT API to generate a question and corresponding answers/options.
• The generated text is then used to create image prompts for Kandinsky.
• Kandinsky generates the images which are then compiled along with static assets (audio and background video) to generate the final video.
Architectural Patterns or Design Principles
• Modular Design: Even though it’s a single script, consider depicting the logical separation between:
• Input generation (GPT module)
• Processing (Kandinsky image generation)
• Media Compilation (video creation using static assets)
• Integration of External Services: Clearly mark the boundaries where external API calls are made.
• Dependency Management: Note that the project uses a requirements.txt file to manage Python libraries, which is essential for reproducibility.
Relevant Technologies and Libraries
• Python as the main language
• OpenAI GPT service for text generation
• Kandinsky API/service for image creation
• Standard media processing libraries (possibly moviepy or similar, although not explicitly mentioned—they might be part of the requirements)
Final Steps
• That each module is distinctly represented.
• Notes/comments if needed (like “Set your API key” from the README).
• Keep the design simple and intuitive, focusing on the linear flow of data through the system.