
Record, convert and stream audio & video
Showing 5 of 52 projects. Click any project card for scope, mentors, and proposal studio.
Mentors: Student: Aditya Banavi
FFmpeg already supports WHIP, allowing users to publish WebRTC streams using HTTP signaling. This project aims to extend WHIP support by adding support for more codecs, enabling multiple media streams, providing full ICE support, and other related improvements.
Mentors: Student: Raja Rathour
Building upon the recently merged asynchronous execution infrastructure in the FFmpeg LibTorch backend, this project targets two high-impact performance improvements: a production-grade zero-copy GPU pipeline and an optional batching engine for throughput scaling. The primary objective is to reduce host-device transfer overhead, improve real-time inference efficiency, and strengthen backend reliability for practical video processing workloads. The project deliverables include implementation, validation, and documentation aligned with FFmpeg quality and integration standards.
Mentors: Student: Nariman Sayed
FFmpeg can send WebRTC streams via WHIP but has no way to receive them. This project fills that gap by implementing a WHEP demuxer in libavformat, built on top of the existing WHIP code. The plan is to reuse SDP, ICE, DTLS, and SRTP as much as possible and just flip the direction. The end goal is making ffplay whep://server/stream work, with a clean patch series ready for ffmpeg-devel.
Mentors: Student: Geo Ster
ProRes is a popular codec for video editors and professional due to its high quality. FFmpeg has supported it for a while, but lacks hardware acceleration outside of MacOS. Using vulkan compute shaders, encoding performance can be increased in a cross platform way, using modern GPUs
Mentors: Student: averne
ProRes is a video codec developed by Apple, and is frequently used as an intermediate in video editing, sitting in-between raw formats and consumer codecs such as H.264. Due to the increasing bandwidth and compute power required to edit high-resolution content (4K and up), a shader-based decoder is desirable to accelerate ProRes processing in video editing pipelines and other situations. This projects aims at delivering a high-performance, Vulkan-based decoder for Apple ProRes, supporting its 4444 and 4444 XQ profiles, and integrated within the ubiquitous FFmpeg framework.