WebRTC Weekly Issue #649
Disruption recovery, sharper pixels, and the right transport for the job
It’s a plumbing week: recovering gracefully when the media backend falls over, keeping session data inside your own VPC, squeezing more perceived quality out of the same bitrate, and picking the right transport for the latency domain you are actually in. The unglamorous engineering is where the real work lives.
⭐ Tsahi’s Pick
Making 1080p screen shares look sharper (Hopp)
Hopp is focusing on remote pair programming, and as such screen sharing is important. Here’s the thing. Screen sharing goes at 4K and above most of the time. If the source is 1080p and you upscale it to 4K, the result isn’t sharp enough. Here, adding a sharpening filter at render time got a visibly crisper image for free. I like this one because it is a reminder that perceived quality in WebRTC is not only a bitrate problem - and upscaling can be improved upon.
📰 This Week in WebRTC
Recovering from service disruptions with the Vonage Video API (Vonage) - Vonage explains their session disruption callbacks, which is an interesting concept
Agentic workflows for WebRTC and real-time development (WebRTC.ventures) - WebRTC.ventures sharing their experience with AI-assisted development in WebRTC
How DynDevice built virtual classrooms into its LMS with OpenVidu (OpenVidu) - a case study of DynDevice LMS choosing OpenVidu to replace video meeting SaaS platforms
Network traffic engineering: WebRTC vs ST 2110 (The Broadcast Bridge) - the failure here for me is the not understanding that managed networks are becoming rarer as time goes by
Your session data shouldn’t leave your VPC by default (rtcStats) - rtcstats-server keeps WebRTC session data inside your own infrastructure; forwarding anything out to rtcstats.com is opt-in only
📣 Sponsor
WebRTC Live is a monthly webinar series featuring real-time industry guests, hosted by WebRTC.ventures.
Join us for the August 12 episode, “Building Robust Voice AI Pipelines for Every Use Case” with Varun Singh of Daily.co / Pipecat.ai.
📢 Releases
libaom v3.14 “Pinkie Pie” - AOMedia’s AV1 reference encoder/decoder ships eight months of encoding and performance improvements since v3.13