PromoGen

PromoGen Logo
PromoGen Website Interface

Inspiration

creating high-quality marketing videos is usually costly, time-consuming, and can seem confusing to many. by utilizing google cloud AI, we aimed to focus on really quick video creation for the use of promoting anyone's brand, making it accessible to businesses and creators to easily generate videos.

What it does

ai-powered marketing video generator that creates dynamic promotional videos tailored to different product categories and social media platforms (Instagram, TikTok, YouTube Shorts). Users provide simple inputs like product name and category, and the AI crafts compelling video prompts, then generates videos automatically via Google’s Gemini API.

How we built it

We built this tool using Python and the Google Gemini API. The core logic includes dynamic prompt generation for video scripts based on product info, platform-specific video configurations, and automation of video generation API calls.

Challenges we ran into

understanding and working within Google Gemini API rate limits and quota management.
crafting prompt templates that produce versatile marketing videos, not limited to narrow product types.
handling asynchronous video generation and polling for completion reliably.
designing the system to accommodate multiple social media platform requirements (aspect ratios, durations). ## Accomplishments that we're proud of
- developed adaptive, category-aware prompt generation to produce realisticmarketing video content. • created a flexible CLI interface for easy generation of videos for various platforms. • integrated polling and reliable video downloading for seamless automation. • designed the system to be extensible for future prompt improvements and additional platforms. ## What we learned
- the importance of highly targeted, clear prompts to use AI-generated content effectively. • managing cloud API quotas and the need for robust error handling and backoff strategies. - learned importance of utilizing simple print statements to check errors in code • optimizing video generation for different social platforms significantly impacts engagement potential. • the limitations of current AI video generation models, particularly around video length and quality. ## What's next for PromoGen
- add support for multi-clip stitching or longer videos to overcome model duration limitations.
- implement advanced prompt tuning and personalization based on customer data.
- build a user-friendly web UI for easier access to video generation features.
  - extend platform support to emerging video social channels.
- incorporate analytics for measuring generated video performance and guiding improvements.
  - produce more accurate and prompt specific video creation
  - visual output with less text error and no "burned" letters -allow user to have access to one of the existing agents to get very specific results