Inspiration

In today’s fast-paced world, everyone loves sharing their stories through short videos on platforms like Instagram and TikTok. However, creating engaging reels still requires time-consuming editing. We wanted to simplify this process while making it more collaborative.

What It Does

With ReelEasy, we built an AI-powered tool that effortlessly selects the best moments from your and your friends’ photos, crafts a compelling story, and generates a professional-quality video, complete with voiceovers and transitions, without any manual editing.

This innovation democratizes video creation, enabling everyone to share their special moments seamlessly, breaking down traditional barriers to digital storytelling.

How We Built It

To bring our AI-powered video generation tool to life, we integrated various powerful libraries and APIs. Here's a breakdown of the key components used in our project:

Libraries/Tools Used

  • Suite Studios – For collecting images from multiple users and accessing them at high speed.
  • MoviePy – For video editing, adding animations, transitions, subtitles, and merging media files.
  • pydub – For audio manipulation, including adjusting volumes and integrating music and voiceovers.
  • Gradio – To build our web interface, allowing users to upload media, provide voice snippets, and download videos.

APIs Used

  • BLIP & MiniLM – For annotating images and selecting the most relevant ones.
  • PlayHT – For cloning the user’s voice and generating realistic voiceovers for personalization.
  • OpenAI – For AI-driven storytelling, generating narratives based on user descriptions.

Challenges We Ran Into

  • Limited expertise in frontend development and its integration with the backend.
  • Dealing with rate limit errors while using an older version of the PlayHT API.
  • Researching and selecting the right combination of models to extract the best images.
  • Ensuring interoperability between Mac and Windows users.
  • Learning and implementing MoviePy for video editing.

Accomplishments That We're Proud Of

  • Successfully building the first iteration of ReelEasy in a short time and addressing unaddressed problems in video creation.
  • Pushing ourselves into unfamiliar domains.
  • Strong collaboration and supporting each other throughout the process.

What We Learned

  • We each brought different expertise but took on roles we hadn’t before. Everything we did, from using Hugging Face models to cloning voices via APIs, was a valuable learning experience.

What's Next for ReelEasy

  • Expanding to support full-length videos.
  • Reducing latency in voiceover generation.
  • Improving cross-platform interoperability.

Built With

  • gradio
  • minilm
  • moviepy
  • openai
  • playht
  • suitestudio
Share this project:

Updates