How nikhilsetiya/prism works
Project type and purpose
• This is a full-stack asset-generation application (CLI + Streamlit UI) that automates the creation of localized marketing assets using an AI image service (OpenAI DALL-E).
• One “brief” describes products, locales, ratios → pipeline → a batch of finished images.
File structure
Top-level files and folders:
• run_prism.py, create_campaign.py, validate_setup.py – entry points for CLI workflows
• app.py + .streamlit/ – Streamlit UI interface
• config.yaml – global configuration for quality, governance, workers, rate limits, post-processing
• src/ – core modules implementing each pipeline stage
– pipeline.py orchestrates the sequence
– governance.py, asset_manager.py, image_generator.py, compositor.py, post_processor.py, storage.py, utils.py, models.py, aspect_ratio_processor.py, localizer.py
• assets/ – brand and input placeholders
• test_campaigns_package/ – sample briefs and data for end-to-end testing
• requirements.txt – Python dependencies
README for additional insights
Brief → Governance → Asset Manager → Generator → Compositor → Post-Process → Output
• Mentions parallel execution, rate limiting, cache-first storage, local (or cloud) storage
• Dual interface: CLI and Streamlit UI