GMCasts
Project contains infrastructure to render videos in frames as well as a new type of frame client which can render video elements
Project Description
The aim of this project is to provide a marketplace for viewing and sharing video streams built with generative AI.
In its final form, each stream cast will have three views. The first view is the fallback with og-image to render in clients which cannot handle video. The first view contains 2 buttons - Preview - publicly available clip preview. Mint- Token mint to be allowed to view the entire clip.
The second view contains - Play button to allow qualified users to play the entire stream. Mint button to allow unqualified users to mint and enter.
The third view contains- Controls for viewing/stopping video and breadcrumbs.
This primary frame view will be composed with other social frames to provide experiences such as communities and channels around content and creators.
How it's Made
The project is primarily built with NextJS React JS middleware/frontend and an expressjs backend.
It is Made extensive use of hub-mono repo and hub-nodejs libraries.
Used Pinata hub for querying and Pinata REST API to populate views.
Copy pasted components (refactored into JS) from coinbase onchainkit and Frames.js Took a lot of inspiration from code-base of both Frames.js and onchainkit while developing my components.
Used livepeer.js extensively for transcoding as well as on the client for playing streams.
cheerio.js for parsing HTML. ffmpeg for streaming video.
Just started implementing karma3labs for some user-interactions (future work)
